aid
stringlengths 9
15
| mid
stringlengths 7
10
| abstract
stringlengths 78
2.56k
| related_work
stringlengths 92
1.77k
| ref_abstract
dict |
|---|---|---|---|---|
1901.09590 | 2914592219 | Knowledge graphs are structured representations of real world facts. However, they typically contain only a small subset of all possible facts. Link prediction is a task of inferring missing facts based on existing ones. We propose TuckER, a relatively simple but powerful linear model based on Tucker decomposition of the binary tensor representation of knowledge graph triples. TuckER outperforms all previous state-of-the-art models across standard link prediction datasets. We prove that TuckER is a fully expressive model, deriving the bound on its entity and relation embedding dimensionality for full expressiveness which is several orders of magnitude smaller than the bound of previous state-of-the-art models ComplEx and SimplE. We further show that several previously introduced linear models can be viewed as special cases of TuckER. | DistMult DistMult @cite_15 is a special case of RESCAL with a diagonal matrix per relation, so the number of parameters of DistMult grows linearly with respect to the embedding dimension, reducing overfitting. However, the linear transformation performed on subject entity embedding vectors in DistMult is limited to a stretch. Given the equivalence of subject and object entity embeddings for the same entity, third-order binary tensor learned by DistMult is symmetric in the subject and object entity mode and thus DistMult cannot model asymmetric relations. | {
"cite_N": [
"@cite_15"
],
"mid": [
"2951077644"
],
"abstract": [
"We consider learning representations of entities and relations in KBs using the neural-embedding approach. We show that most existing models, including NTN (, 2013) and TransE (, 2013b), can be generalized under a unified learning framework, where entities are low-dimensional vectors learned from a neural network and relations are bilinear and or linear mapping functions. Under this framework, we compare a variety of embedding models on the link prediction task. We show that a simple bilinear formulation achieves new state-of-the-art results for the task (achieving a top-10 accuracy of 73.2 vs. 54.7 by TransE on Freebase). Furthermore, we introduce a novel approach that utilizes the learned relation embeddings to mine logical rules such as \"BornInCity(a,b) and CityInCountry(b,c) => Nationality(a,c)\". We find that embeddings learned from the bilinear objective are particularly good at capturing relational semantics and that the composition of relations is characterized by matrix multiplication. More interestingly, we demonstrate that our embedding-based rule extraction approach successfully outperforms a state-of-the-art confidence-based rule mining approach in mining Horn rules that involve compositional reasoning."
]
} |
1901.09590 | 2914592219 | Knowledge graphs are structured representations of real world facts. However, they typically contain only a small subset of all possible facts. Link prediction is a task of inferring missing facts based on existing ones. We propose TuckER, a relatively simple but powerful linear model based on Tucker decomposition of the binary tensor representation of knowledge graph triples. TuckER outperforms all previous state-of-the-art models across standard link prediction datasets. We prove that TuckER is a fully expressive model, deriving the bound on its entity and relation embedding dimensionality for full expressiveness which is several orders of magnitude smaller than the bound of previous state-of-the-art models ComplEx and SimplE. We further show that several previously introduced linear models can be viewed as special cases of TuckER. | ComplEx ComplEx @cite_1 extends DistMult to the complex domain. Even though each relation matrix of ComplEx is still diagonal, subject and object entity embeddings for the same entity are no longer equivalent, but complex conjugates, which introduces asymmetry into the tensor decomposition and thus enables ComplEx to model asymmetric relations. | {
"cite_N": [
"@cite_1"
],
"mid": [
"2942026896"
],
"abstract": [
"In this work, we move beyond the traditional complex-valued representations, introducing more expressive hypercomplex representations to model entities and relations for knowledge graph embeddings. More specifically, quaternion embeddings, hypercomplex-valued embeddings with three imaginary components, are utilized to represent entities. Relations are modelled as rotations in the quaternion space. The advantages of the proposed approach are: (1) Latent inter-dependencies (between all components) are aptly captured with Hamilton product, encouraging a more compact interaction between entities and relations; (2) Quaternions enable expressive rotation in four-dimensional space and have more degree of freedom than rotation in complex plane; (3) The proposed framework is a generalization of ComplEx on hypercomplex space while offering better geometrical interpretations, concurrently satisfying the key desiderata of relational representation learning (i.e., modeling symmetry, anti-symmetry and inversion). Experimental results demonstrate that our method achieves state-of-the-art performance on four well-established knowledge graph completion benchmarks."
]
} |
1901.09590 | 2914592219 | Knowledge graphs are structured representations of real world facts. However, they typically contain only a small subset of all possible facts. Link prediction is a task of inferring missing facts based on existing ones. We propose TuckER, a relatively simple but powerful linear model based on Tucker decomposition of the binary tensor representation of knowledge graph triples. TuckER outperforms all previous state-of-the-art models across standard link prediction datasets. We prove that TuckER is a fully expressive model, deriving the bound on its entity and relation embedding dimensionality for full expressiveness which is several orders of magnitude smaller than the bound of previous state-of-the-art models ComplEx and SimplE. We further show that several previously introduced linear models can be viewed as special cases of TuckER. | ConvE ConvE @cite_32 is the first non-linear model that significantly outperformed the preceding linear models. In ConvE, a global 2D convolution operation is performed on the subject entity and relation embedding vectors, after they are reshaped to matrices and concatenated. The obtained feature maps are flattened, transformed through a fully connected layer, and the inner product is taken with all object entity vectors to generate a score for each triple. Whilst results achieved by ConvE are impressive, its reshaping and concatenating of vectors as well as using 2D convolution on word embeddings is unintuitive. | {
"cite_N": [
"@cite_32"
],
"mid": [
"2774837955"
],
"abstract": [
"We introduce a novel embedding method for knowledge base completion task. Our approach advances state-of-the-art (SOTA) by employing a convolutional neural network (CNN) for the task which can capture global relationships and transitional characteristics. We represent each triple (head entity, relation, tail entity) as a 3-column matrix which is the input for the convolution layer. Different filters having a same shape of 1x3 are operated over the input matrix to produce different feature maps which are then concatenated into a single feature vector. This vector is used to return a score for the triple via a dot product. The returned score is used to predict whether the triple is valid or not. Experiments show that ConvKB achieves better link prediction results than previous SOTA models on two current benchmark datasets WN18RR and FB15k-237."
]
} |
1901.09590 | 2914592219 | Knowledge graphs are structured representations of real world facts. However, they typically contain only a small subset of all possible facts. Link prediction is a task of inferring missing facts based on existing ones. We propose TuckER, a relatively simple but powerful linear model based on Tucker decomposition of the binary tensor representation of knowledge graph triples. TuckER outperforms all previous state-of-the-art models across standard link prediction datasets. We prove that TuckER is a fully expressive model, deriving the bound on its entity and relation embedding dimensionality for full expressiveness which is several orders of magnitude smaller than the bound of previous state-of-the-art models ComplEx and SimplE. We further show that several previously introduced linear models can be viewed as special cases of TuckER. | HypER HypER @cite_9 is a simplified convolutional model, that uses a hypernetwork to generate 1D convolutional filters for each relation, extracting relation-specific features from subject entity embeddings. The authors show that convolution is a way of introducing sparsity and parameter tying and that HypER can be understood in terms of tensor factorization up to a non-linearity, thus placing HypER closer to the well established family of factorization models. The drawback of HypER is that it sets most elements of the core weight tensor to 0, which amounts to hard regularization, rather than letting the model learn which parameters to use via a soft regularization approach. | {
"cite_N": [
"@cite_9"
],
"mid": [
"2127409454"
],
"abstract": [
"Beyond linear and kernel-based feature extraction, we propose in this paper the generalized feature extraction formulation based on the so-called graph embedding framework. Two novel correlation metric based algorithms are presented based on this formulation. correlation embedding analysis (CEA), which incorporates both correlational mapping and discriminating analysis, boosts the discriminating power by mapping data from a high-dimensional hypersphere onto another low-dimensional hypersphere and preserving the intrinsic neighbor relations with local graph modeling. correlational principal component analysis (CPCA) generalizes the conventional Principal Component Analysis (PCA) algorithm to the case with data distributed on a high-dimensional hypersphere. Their advantages stem from two facts: 1) tailored to normalized data, which are often the outputs from the data preprocessing step, and 2) directly designed with correlation metric, which shows to be generally better than Euclidean distance for classification purpose. Extensive comparisons with existing algorithms on visual classification experiments demonstrate the effectiveness of the proposed algorithms."
]
} |
1901.09891 | 2913012226 | Data augmentation is usually adopted to increase the amount of training data, prevent overfitting and improve the performance of deep models. However, in practice, random data augmentation, such as random image cropping, is low-efficiency and might introduce many uncontrolled background noises. In this paper, we propose Weakly Supervised Data Augmentation Network (WS-DAN) to explore the potential of data augmentation. Specifically, for each training image, we first generate attention maps to represent the object's discriminative parts by weakly supervised learning. Next, we augment the image guided by these attention maps, including attention cropping and attention dropping. The proposed WS-DAN improves the classification accuracy in two folds. In the first stage, images can be seen better since more discriminative parts' features will be extracted. In the second stage, attention regions provide accurate location of object, which ensures our model to look at the object closer and further improve the performance. Comprehensive experiments in common fine-grained visual classification datasets show that our WS-DAN surpasses the state-of-the-art methods, which demonstrates its effectiveness. | Random data augmentation suffers from low efficiency and generating much uncontrolled noise data. To overcome these issue, a few methods have been proposed to take dataset distribution into consideration and augment data according to the feedback of the training dataset, which is more effective than a random distribution. Cubuk al proposed AutoAugmentation @cite_16 to create a search space of data augmentation policies. It can automatically design a specific policy so as to obtain state-of-the-art validation accuracy for target dataset. Peng al proposed Adversarial Data Augmentation @cite_43 to jointly optimize data augmentation and deep model. They designed an augmentation network to online generate data and improve the robustness of the deep model. However, their data specific augmentation designing process is significantly complicated than random augmentation. Our attention guided data augmentation is more simple and direct, which can be easily trained end-to-end. | {
"cite_N": [
"@cite_43",
"@cite_16"
],
"mid": [
"2963468256",
"2798409409"
],
"abstract": [
"Random data augmentation is a critical technique to avoid overfitting in training deep models. Yet, data augmentation and network training are often two isolated processes in most settings, yielding to a suboptimal training. Why not jointly optimize the two? We propose adversarial data augmentation to address this limitation. The key idea is to design a generator (e.g. an augmentation network) that competes against a discriminator (e.g. a target network) by generating hard examples online. The generator explores weaknesses of the discriminator, while the discriminator learns from hard augmentations to achieve better performance. A reward penalty strategy is also proposed for efficient joint training. We investigate human pose estimation and carry out comprehensive ablation studies to validate our method. The results prove that our method can effectively improve state-of-the-art models without additional data effort.",
"Random data augmentation is a critical technique to avoid overfitting in training deep neural network models. However, data augmentation and network training are usually treated as two isolated processes, limiting the effectiveness of network training. Why not jointly optimize the two? We propose adversarial data augmentation to address this limitation. The main idea is to design an augmentation network (generator) that competes against a target network (discriminator) by generating hard' augmentation operations online. The augmentation network explores the weaknesses of the target network, while the latter learns from hard' augmentations to achieve better performance. We also design a reward penalty strategy for effective joint training. We demonstrate our approach on the problem of human pose estimation and carry out a comprehensive experimental analysis, showing that our method can significantly improve state-of-the-art models without additional data efforts."
]
} |
1901.09891 | 2913012226 | Data augmentation is usually adopted to increase the amount of training data, prevent overfitting and improve the performance of deep models. However, in practice, random data augmentation, such as random image cropping, is low-efficiency and might introduce many uncontrolled background noises. In this paper, we propose Weakly Supervised Data Augmentation Network (WS-DAN) to explore the potential of data augmentation. Specifically, for each training image, we first generate attention maps to represent the object's discriminative parts by weakly supervised learning. Next, we augment the image guided by these attention maps, including attention cropping and attention dropping. The proposed WS-DAN improves the classification accuracy in two folds. In the first stage, images can be seen better since more discriminative parts' features will be extracted. In the second stage, attention regions provide accurate location of object, which ensures our model to look at the object closer and further improve the performance. Comprehensive experiments in common fine-grained visual classification datasets show that our WS-DAN surpasses the state-of-the-art methods, which demonstrates its effectiveness. | To focus on the local features, many methods rely on the annotations of parts location or attribute. Part R-CNN @cite_44 extended R-CNN @cite_32 to detect objects and localize their parts under a geometric prior, then predicted a fine-grained category from a pose-normalized representation. @cite_33 proposed a feedback-control framework Deep LAC to back-propagate alignment and classification errors to localization; they also proposed a valve linkage function (VLF) to connect the localization and classification modules. | {
"cite_N": [
"@cite_44",
"@cite_32",
"@cite_33"
],
"mid": [
"2560096627",
"2950461853",
"2773003563"
],
"abstract": [
"Deep convolution neural networks (CNNs) have demonstrated advanced performance on single-label image classification, and various progress also has been made to apply CNN methods on multilabel image classification, which requires annotating objects, attributes, scene categories, etc., in a single shot. Recent state-of-the-art approaches to the multilabel image classification exploit the label dependencies in an image, at the global level, largely improving the labeling capacity. However, predicting small objects and visual concepts is still challenging due to the limited discrimination of the global visual features. In this paper, we propose a regional latent semantic dependencies model (RLSD) to address this problem. The utilized model includes a fully convolutional localization architecture to localize the regions that may contain multiple highly dependent labels. The localized regions are further sent to the recurrent neural networks to characterize the latent semantic dependencies at the regional level. Experimental results on several benchmark datasets show that our proposed model achieves the best performance compared to the state-of-the-art models, especially for predicting small objects occurring in the images. Also, we set up an upper bound model (RLSD+ft-RPN) using bounding-box coordinates during training, and the experimental results also show that our RLSD can approach the upper bound without using the bounding-box annotations, which is more realistic in the real world.",
"Deep convolution neural networks (CNN) have demonstrated advanced performance on single-label image classification, and various progress also have been made to apply CNN methods on multi-label image classification, which requires to annotate objects, attributes, scene categories etc. in a single shot. Recent state-of-the-art approaches to multi-label image classification exploit the label dependencies in an image, at global level, largely improving the labeling capacity. However, predicting small objects and visual concepts is still challenging due to the limited discrimination of the global visual features. In this paper, we propose a Regional Latent Semantic Dependencies model (RLSD) to address this problem. The utilized model includes a fully convolutional localization architecture to localize the regions that may contain multiple highly-dependent labels. The localized regions are further sent to the recurrent neural networks (RNN) to characterize the latent semantic dependencies at the regional level. Experimental results on several benchmark datasets show that our proposed model achieves the best performance compared to the state-of-the-art models, especially for predicting small objects occurred in the images. In addition, we set up an upper bound model (RLSD+ft-RPN) using bounding box coordinates during training, the experimental results also show that our RLSD can approach the upper bound without using the bounding-box annotations, which is more realistic in the real world.",
"Recognizing fine-grained categories (e.g., bird species) highly relies on discriminative part localization and part-based fine-grained feature learning. Existing approaches predominantly solve these challenges independently, while neglecting the fact that part localization (e.g., head of a bird) and fine-grained feature learning (e.g., head shape) are mutually correlated. In this paper, we propose a novel part learning approach by a multi-attention convolutional neural network (MA-CNN), where part generation and feature learning can reinforce each other. MA-CNN consists of convolution, channel grouping and part classification sub-networks. The channel grouping network takes as input feature channels from convolutional layers, and generates multiple parts by clustering, weighting and pooling from spatially-correlated channels. The part classification network further classifies an image by each individual part, through which more discriminative fine-grained features can be learned. Two losses are proposed to guide the multi-task learning of channel grouping and part classification, which encourages MA-CNN to generate more discriminative parts from feature channels and learn better fine-grained features from parts in a mutual reinforced way. MA-CNN does not need bounding box part annotation and can be trained end-to-end. We incorporate the learned parts from MA-CNN with part-CNN for recognition, and show the best performances on three challenging published fine-grained datasets, e.g., CUB-Birds, FGVC-Aircraft and Stanford-Cars."
]
} |
1901.09891 | 2913012226 | Data augmentation is usually adopted to increase the amount of training data, prevent overfitting and improve the performance of deep models. However, in practice, random data augmentation, such as random image cropping, is low-efficiency and might introduce many uncontrolled background noises. In this paper, we propose Weakly Supervised Data Augmentation Network (WS-DAN) to explore the potential of data augmentation. Specifically, for each training image, we first generate attention maps to represent the object's discriminative parts by weakly supervised learning. Next, we augment the image guided by these attention maps, including attention cropping and attention dropping. The proposed WS-DAN improves the classification accuracy in two folds. In the first stage, images can be seen better since more discriminative parts' features will be extracted. In the second stage, attention regions provide accurate location of object, which ensures our model to look at the object closer and further improve the performance. Comprehensive experiments in common fine-grained visual classification datasets show that our WS-DAN surpasses the state-of-the-art methods, which demonstrates its effectiveness. | Weakly supervised learning is an umbrella term that covers a variety of studies that attempt to construct predictive models by learning with weak supervision @cite_0 , which mainly consists of incomplete, inexact and inaccurate supervision. Localizing object or its parts only by image-level annotation belongs to the inexact supervision. | {
"cite_N": [
"@cite_0"
],
"mid": [
"1934621328"
],
"abstract": [
"Weakly supervised object detection, is a challenging task, where the training procedure involves learning at the same time both, the model appearance and the object location in each image. The classical approach to solve this problem is to consider the location of the object of interest in each image as a latent variable and minimize the loss generated by such latent variable during learning. However, as learning appearance and localization are two interconnected tasks, the optimization is not convex and the procedure can easily get stuck in a poor local minimum, i.e. the algorithm “misses” the object in some images. In this paper, we help the optimization to get close to the global minimum by enforcing a “soft” similarity between each possible location in the image and a reduced set of “exemplars”, or clusters, learned with a convex formulation in the training images. The help is effective because it comes from a different and smooth source of information that is not directly connected with the main task. Results show that our method improves a strong baseline based on convolutional neural network features by more than 4 points without any additional features or extra computation at testing time but only adding a small increment of the training time due to the convex clustering."
]
} |
1901.09891 | 2913012226 | Data augmentation is usually adopted to increase the amount of training data, prevent overfitting and improve the performance of deep models. However, in practice, random data augmentation, such as random image cropping, is low-efficiency and might introduce many uncontrolled background noises. In this paper, we propose Weakly Supervised Data Augmentation Network (WS-DAN) to explore the potential of data augmentation. Specifically, for each training image, we first generate attention maps to represent the object's discriminative parts by weakly supervised learning. Next, we augment the image guided by these attention maps, including attention cropping and attention dropping. The proposed WS-DAN improves the classification accuracy in two folds. In the first stage, images can be seen better since more discriminative parts' features will be extracted. In the second stage, attention regions provide accurate location of object, which ensures our model to look at the object closer and further improve the performance. Comprehensive experiments in common fine-grained visual classification datasets show that our WS-DAN surpasses the state-of-the-art methods, which demonstrates its effectiveness. | Accurately locating the object or its parts only by image-level supervision is very challenging. Early work @cite_7 @cite_13 usually generate class-specific localization maps by Global Average Pooling (GAP) @cite_3 . The activation area can reflect the location of an object. However, training by softmax cross entropy loss usually leads the model to pay attention to the most discriminative location, whose output bounding box just covers part of the object. To locate the whole object. Singh al @cite_22 randomly hides the patches of input images so as to force the network to find other discriminative parts. However, the process is inefficient for the lack of high-level guidance. Zhang al proposed Adversarial Complementary Learning (ACoL) @cite_4 approach to discover entire objects by training two adversary complementary classifiers, which can locate different object parts and discover the complementary regions that belong to the same object. Nevertheless, there are only two complementary regions in their implementation, which limits accuracy. Our attention-guided data augmentation encourages the model to pay attention to multiple object parts, extract more discriminative features and achieve better performance in object localizing. | {
"cite_N": [
"@cite_4",
"@cite_22",
"@cite_7",
"@cite_3",
"@cite_13"
],
"mid": [
"2937970997",
"2964274719",
"2543332268",
"2441255125",
"1919709169"
],
"abstract": [
"Many state-of-the-art approaches for object recognition reduce the problem to a 0-1 classification task. This allows one to leverage sophisticated machine learning techniques for training classifiers from labeled examples. However, these models are typically trained independently for each class using positive and negative examples cropped from images. At test-time, various post-processing heuristics such as non-maxima suppression (NMS) are required to reconcile multiple detections within and between different classes for each image. Though crucial to good performance on benchmarks, this post-processing is usually defined heuristically. We introduce a unified model for multi-class object recognition that casts the problem as a structured prediction task. Rather than predicting a binary label for each image window independently, our model simultaneously predicts a structured labeling of the entire image (Fig. 1). Our model learns statistics that capture the spatial arrangements of various object classes in real images, both in terms of which arrangements to suppress through NMS and which arrangements to favor through spatial co-occurrence statistics. We formulate parameter estimation in our model as a max-margin learning problem. Given training images with ground-truth object locations, we show how to formulate learning as a convex optimization problem. We employ the cutting plane algorithm of (Mach. Learn. 2009) to efficiently learn a model from thousands of training images. We show state-of-the-art results on the PASCAL VOC benchmark that indicate the benefits of learning a global model encapsulating the spatial layout of multiple object classes (a preliminary version of this work appeared in ICCV 2009, , IEEE international conference on computer vision, 2009).",
"In this work, we propose Adversarial Complementary Learning (ACoL) to automatically localize integral objects of semantic interest with weak supervision. We first mathematically prove that class localization maps can be obtained by directly selecting the class-specific feature maps of the last convolutional layer, which paves a simple way to identify object regions. We then present a simple network architecture including two parallel-classifiers for object localization. Specifically, we leverage one classification branch to dynamically localize some discriminative object regions during the forward pass. Although it is usually responsive to sparse parts of the target objects, this classifier can drive the counterpart classifier to discover new and complementary object regions by erasing its discovered regions from the feature maps. With such an adversarial learning, the two parallel-classifiers are forced to leverage complementary object regions for classification and can finally generate integral object localization together. The merits of ACoL are mainly two-fold: 1) it can be trained in an end-to-end manner; 2) dynamically erasing enables the counterpart classifier to discover complementary object regions more effectively. We demonstrate the superiority of our ACoL approach in a variety of experiments. In particular, the Top-1 localization error rate on the ILSVRC dataset is 45.14 , which is the new state-of-the-art.",
"This paper present a part-based approach for detecting objects with large variation of appearance. We extract local image patches as local features both from the object and from the background in training images to learn an object part model discriminatively. Our object part model discriminates the local features whether they are an object part or not. Based on the discrimination results, each local feature casts probabilistic votes for the object location and size which are learned from the training images. Our object part model also requires regression performance for predicting the object location and size through the voting procedure. We build such an object part model with an ensemble of randomized trees trained by splitting each tree node so as to reduce the entropy of class label distribution and the variance of object location and size. Experimental results on hand detection with large pose variation show that our approach outperforms conventional generalized Hough transform. We verified the performance on a public dataset of side-view cars.",
"We address the problem of weakly supervised object localization where only image-level annotations are available for training. Many existing approaches tackle this problem through object proposal mining. However, a substantial amount of noise in object proposals causes ambiguities for learning discriminative object models. Such approaches are sensitive to model initialization and often converge to an undesirable local minimum. In this paper, we address this problem by progressive domain adaptation with two main steps: classification adaptation and detection adaptation. In classification adaptation, we transfer a pre-trained network to our multi-label classification task for recognizing the presence of a certain object in an image. In detection adaptation, we first use a mask-out strategy to collect class-specific object proposals and apply multiple instance learning to mine confident candidates. We then use these selected object proposals to fine-tune all the layers, resulting in a fully adapted detection network. We extensively evaluate the localization performance on the PASCAL VOC and ILSVRC datasets and demonstrate significant performance improvement over the state-of-the-art methods.",
"This paper addresses unsupervised discovery and localization of dominant objects from a noisy image collection with multiple object classes. The setting of this problem is fully unsupervised, without even image-level annotations or any assumption of a single dominant class. This is far more general than typical colocalization, cosegmentation, or weakly-supervised localization tasks. We tackle the discovery and localization problem using a part-based region matching approach: We use off-the-shelf region proposals to form a set of candidate bounding boxes for objects and object parts. These regions are efficiently matched across images using a probabilistic Hough transform that evaluates the confidence for each candidate correspondence considering both appearance and spatial consistency. Dominant objects are discovered and localized by comparing the scores of candidate regions and selecting those that stand out over other regions containing them. Extensive experimental evaluations on standard benchmarks demonstrate that the proposed approach significantly outperforms the current state of the art in colocalization, and achieves robust object discovery in challenging mixed-class datasets."
]
} |
1901.09888 | 2913777072 | The increasing interest in user privacy is leading to new privacy preserving machine learning paradigms. In the Federated Learning paradigm, a master machine learning model is distributed to user clients, the clients use their locally stored data and model for both inference and calculating model updates. The model updates are sent back and aggregated on the server to update the master model then redistributed to the clients. In this paradigm, the user data never leaves the client, greatly enhancing the user' privacy, in contrast to the traditional paradigm of collecting, storing and processing user data on a backend server beyond the user's control. In this paper we introduce, as far as we are aware, the first federated implementation of a Collaborative Filter. The federated updates to the model are based on a stochastic gradient approach. As a classical case study in machine learning, we explore a personalized recommendation system based on users' implicit feedback and demonstrate the method's applicability to both the MovieLens and an in-house dataset. Empirical validation confirms a collaborative filter can be federated without a loss of accuracy compared to a standard implementation, hence enhancing the user's privacy in a widely used recommender application while maintaining recommender performance. | This work lies at the intersection of three research topics: (i) matrix factorization, (ii) parallel & distributed and, (iii) federated learning. Recently, Alternating Least Squares (ALS) and Stochastic Gradient Descent (SGD) have gained much interest and have become the most popular algorithms for matrix factorization in recommender systems @cite_25 . The ALS algorithm allows learning the latent factor matrices by alternating between updates to one factor matrix while holding the other factor matrix fixed. Each iteration of an update to the latent factor matrices is referred to as an . Although the time complexity per epoch is cubic in the number of factors, numerous studies show the ALS is well suited for parallelization @cite_13 @cite_12 @cite_9 @cite_10 @cite_4 . It is not merely a coincidence that ALS is the premiere parallel matrix factorization implementation for CF in Apache Spark ( https: spark.apache.org docs latest mllib-collaborative-filtering.html). | {
"cite_N": [
"@cite_13",
"@cite_4",
"@cite_9",
"@cite_10",
"@cite_25",
"@cite_12"
],
"mid": [
"2020098476",
"2142466236",
"2952647294",
"2029463952",
"2964341462",
"2963626304"
],
"abstract": [
"Matrix factorization, when the matrix has missing values, has become one of the leading techniques for recommender systems. To handle web-scale datasets with millions of users and billions of ratings, scalability becomes an important issue. Alternating Least Squares (ALS) and Stochastic Gradient Descent (SGD) are two popular approaches to compute matrix factorization. There has been a recent flurry of activity to parallelize these algorithms. However, due to the cubic time complexity in the target rank, ALS is not scalable to large-scale datasets. On the other hand, SGD conducts efficient updates but usually suffers from slow convergence that is sensitive to the parameters. Coordinate descent, a classical optimization approach, has been used for many other large-scale problems, but its application to matrix factorization for recommender systems has not been explored thoroughly. In this paper, we show that coordinate descent based methods have a more efficient update rule compared to ALS, and are faster and have more stable convergence than SGD. We study different update sequences and propose the CCD++ algorithm, which updatesrank-one factors one by one. In addition, CCD++ can be easily parallelized on both multi-core and distributed systems. We empirically show that CCD++ is much faster than ALS and SGD in both settings. As an example, on a synthetic dataset with 2 billion ratings, CCD++ is 4 times faster than both SGD and ALS using a distributed system with 20 machines.",
"Matrix factorization, when the matrix has missing values, has become one of the leading techniques for recommender systems. To handle web-scale datasets with millions of users and billions of ratings, scalability becomes an important issue. Alternating least squares (ALS) and stochastic gradient descent (SGD) are two popular approaches to compute matrix factorization, and there has been a recent flurry of activity to parallelize these algorithms. However, due to the cubic time complexity in the target rank, ALS is not scalable to large-scale datasets. On the other hand, SGD conducts efficient updates but usually suffers from slow convergence that is sensitive to the parameters. Coordinate descent, a classical optimization approach, has been used for many other large-scale problems, but its application to matrix factorization for recommender systems has not been thoroughly explored. In this paper, we show that coordinate descent-based methods have a more efficient update rule compared to ALS and have faster and more stable convergence than SGD. We study different update sequences and propose the CCD++ algorithm, which updates rank-one factors one by one. In addition, CCD++ can be easily parallelized on both multi-core and distributed systems. We empirically show that CCD++ is much faster than ALS and SGD in both settings. As an example, with a synthetic dataset containing 14.6 billion ratings, on a distributed memory cluster with 64 processors, to deliver the desired test RMSE, CCD++ is 49 times faster than SGD and 20 times faster than ALS. When the number of processors is increased to 256, CCD++ takes only 16 s and is still 40 times faster than SGD and 20 times faster than ALS.",
"We present a technique for significantly speeding up Alternating Least Squares (ALS) and Gradient Descent (GD), two widely used algorithms for tensor factorization. By exploiting properties of the Khatri-Rao product, we show how to efficiently address a computationally challenging sub-step of both algorithms. Our algorithm, DFacTo, only requires two sparse matrix-vector products and is easy to parallelize. DFacTo is not only scalable but also on average 4 to 10 times faster than competing algorithms on a variety of datasets. For instance, DFacTo only takes 480 seconds on 4 machines to perform one iteration of the ALS algorithm and 1,143 seconds to perform one iteration of the GD algorithm on a 6.5 million x 2.5 million x 1.5 million dimensional tensor with 1.2 billion non-zero entries.",
"The efficient, distributed factorization of large matrices on clusters of commodity machines is crucial to applying latent factor models in industrial-scale recommender systems. We propose an efficient, data-parallel low-rank matrix factorization with Alternating Least Squares which uses a series of broadcast-joins that can be efficiently executed with MapReduce. We empirically show that the performance of our solution is suitable for real-world use cases. We present experiments on two publicly available datasets and on a synthetic dataset termed Bigflix, generated from the Netflix dataset. Bigflix contains 25 million users and more than 5 billion ratings, mimicking data sizes recently reported as Netflix' production workload. We demonstrate that our approach is able to run an iteration of Alternating Least Squares in six minutes on this dataset. Our implementation has been contributed to the open source machine learning library Apache Mahout.",
"This paper introduces a randomized variation of the alternating least squares (ALS) algorithm for rank reduction of canonical tensor formats. The aim is to address the potential numerical ill-conditioning of least squares matrices at each ALS iteration. The proposed algorithm, dubbed randomized ALS, mitigates large condition numbers via projections onto random tensors, a technique inspired by well-established randomized projection methods for solving overdetermined least squares problems in a matrix setting. A probabilistic bound on the condition numbers of the randomized ALS matrices is provided, demonstrating reductions relative to their standard counterparts. Additionally, results are provided that guarantee comparable accuracy of the randomized ALS solution at each iteration. The performance of the randomized algorithm is studied with three examples, including manufactured tensors and an elliptic PDE with random inputs. In particular, for the latter, tests illustrate not only improvements in condition...",
"Summary Alternating least squares (ALS) is often considered the workhorse algorithm for computing the rank-R canonical tensor approximation, but for certain problems, its convergence can be very slow. The nonlinear conjugate gradient (NCG) method was recently proposed as an alternative to ALS, but the results indicated that NCG is usually not faster than ALS. To improve the convergence speed of NCG, we consider a nonlinearly preconditioned NCG (PNCG) algorithm for computing the rank-R canonical tensor decomposition. Our approach uses ALS as a nonlinear preconditioner in the NCG algorithm. Alternatively, NCG can be viewed as an acceleration process for ALS. We demonstrate numerically that the convergence acceleration mechanism in PNCG often leads to important pay-offs for difficult tensor decomposition problems, with convergence that is significantly faster and more robust than for the stand-alone NCG or ALS algorithms. We consider several approaches for incorporating the nonlinear preconditioner into the NCG algorithm that have been described in the literature previously and have met with success in certain application areas. However, it appears that the nonlinearly PNCG approach has received relatively little attention in the broader community and remains underexplored both theoretically and experimentally. Thus, this paper serves several additional functions, by providing in one place a concise overview of several PNCG variants and their properties that have only been described in a few places scattered throughout the literature, by systematically comparing the performance of these PNCG variants for the tensor decomposition problem, and by drawing further attention to the usefulness of nonlinearly PNCG as a general tool. In addition, we briefly discuss the convergence of the PNCG algorithm. In particular, we obtain a new convergence result for one of the PNCG variants under suitable conditions, building on known convergence results for non-preconditioned NCG. Copyright © 2014 John Wiley & Sons, Ltd."
]
} |
1901.09888 | 2913777072 | The increasing interest in user privacy is leading to new privacy preserving machine learning paradigms. In the Federated Learning paradigm, a master machine learning model is distributed to user clients, the clients use their locally stored data and model for both inference and calculating model updates. The model updates are sent back and aggregated on the server to update the master model then redistributed to the clients. In this paradigm, the user data never leaves the client, greatly enhancing the user' privacy, in contrast to the traditional paradigm of collecting, storing and processing user data on a backend server beyond the user's control. In this paper we introduce, as far as we are aware, the first federated implementation of a Collaborative Filter. The federated updates to the model are based on a stochastic gradient approach. As a classical case study in machine learning, we explore a personalized recommendation system based on users' implicit feedback and demonstrate the method's applicability to both the MovieLens and an in-house dataset. Empirical validation confirms a collaborative filter can be federated without a loss of accuracy compared to a standard implementation, hence enhancing the user's privacy in a widely used recommender application while maintaining recommender performance. | Federated Learning, on the other hand, a distributed learning paradigm essentially assumes user data is not available on central servers and is private and confidential. A prominent direction of research in this domain is based on the weighted averaging of the model parameters @cite_17 @cite_24 . In practice, a master machine learning model is distributed to user clients. Each client updates the local copy of the model weights using the user's personal data and sends updated weights to the server which uses the weighted average of the clients local model weights to update the master model. This federated averaging approach has recently attracted much attention for deep neural networks, however, the same approach may not be applicable to a wide class of other machine learning models such as matrix factorization. Classical studies based on federated averaging used CNNs to train on benchmark image recognition tasks @cite_17 , and LSTM on a language modeling tasks @cite_2 @cite_15 . As a follow-up analysis on federating deep learning models, numerous studies have been proposed addressing the optimization of the communication payloads, noisy, unbalanced @cite_23 , non-iid and massively distributed data @cite_33 . | {
"cite_N": [
"@cite_33",
"@cite_24",
"@cite_23",
"@cite_2",
"@cite_15",
"@cite_17"
],
"mid": [
"2900120080",
"2807006176",
"2283463896",
"2903471046",
"2535838896",
"2900594532"
],
"abstract": [
"We train a recurrent neural network language model using a distributed, on-device learning framework called federated learning for the purpose of next-word prediction in a virtual keyboard for smartphones. Server-based training using stochastic gradient descent is compared with training on client devices using the Federated Averaging algorithm. The federated algorithm, which enables training on a higher-quality dataset for this use case, is shown to achieve better prediction recall. This work demonstrates the feasibility and benefit of training language models on client devices without exporting sensitive user data to servers. The federated learning environment gives users greater control over the use of their data and simplifies the task of incorporating privacy by default with distributed training and aggregation across a population of client devices.",
"Federated learning enables resource-constrained edge compute devices, such as mobile phones and IoT devices, to learn a shared model for prediction, while keeping the training data local. This decentralized approach to train models provides privacy, security, regulatory and economic benefits. In this work, we focus on the statistical challenge of federated learning when local data is non-IID. We first show that the accuracy of federated learning reduces significantly, by up to 55 for neural networks trained for highly skewed non-IID data, where each client device trains only on a single class of data. We further show that this accuracy reduction can be explained by the weight divergence, which can be quantified by the earth mover's distance (EMD) between the distribution over classes on each device and the population distribution. As a solution, we propose a strategy to improve training on non-IID data by creating a small subset of data which is globally shared between all the edge devices. Experiments show that accuracy can be increased by 30 for the CIFAR-10 dataset with only 5 globally shared data.",
"Modern mobile devices have access to a wealth of data suitable for learning models, which in turn can greatly improve the user experience on the device. For example, language models can improve speech recognition and text entry, and image models can automatically select good photos. However, this rich data is often privacy sensitive, large in quantity, or both, which may preclude logging to the data-center and training there using conventional approaches. We advocate an alternative that leaves the training data distributed on the mobile devices, and learns a shared model by aggregating locally-computed updates. We term this decentralized approach Federated Learning. We present a practical method for the federated learning of deep networks that proves robust to the unbalanced and non-IID data distributions that naturally arise. This method allows high-quality models to be trained in relatively few rounds of communication, the principal constraint for federated learning. The key insight is that despite the non-convex loss functions we optimize, parameter averaging over updates from multiple clients produces surprisingly good results, for example decreasing the communication needed to train an LSTM language model by two orders of magnitude.",
"On-device machine learning (ML) enables the training process to exploit a massive amount of user-generated private data samples. To enjoy this benefit, inter-device communication overhead should be minimized. With this end, we propose federated distillation (FD), a distributed model training algorithm whose communication payload size is much smaller than a benchmark scheme, federated learning (FL), particularly when the model size is large. Moreover, user-generated data samples are likely to become non-IID across devices, which commonly degrades the performance compared to the case with an IID dataset. To cope with this, we propose federated augmentation (FAug), where each device collectively trains a generative model, and thereby augments its local data towards yielding an IID dataset. Empirical studies demonstrate that FD with FAug yields around 26x less communication overhead while achieving 95-98 test accuracy compared to FL.",
"Federated Learning is a machine learning setting where the goal is to train a high-quality centralized model while training data remains distributed over a large number of clients each with unreliable and relatively slow network connections. We consider learning algorithms for this setting where on each round, each client independently computes an update to the current model based on its local data, and communicates this update to a central server, where the client-side updates are aggregated to compute a new global model. The typical clients in this setting are mobile phones, and communication efficiency is of the utmost importance. In this paper, we propose two ways to reduce the uplink communication costs: structured updates, where we directly learn an update from a restricted space parametrized using a smaller number of variables, e.g. either low-rank or a random mask; and sketched updates, where we learn a full model update and then compress it using a combination of quantization, random rotations, and subsampling before sending it to the server. Experiments on both convolutional and recurrent networks show that the proposed methods can reduce the communication cost by two orders of magnitude.",
"Federated learning is an approach to distributed machine learning where a global model is learned by aggregating models that have been trained locally on data-generating clients. Contrary to centralized optimization, clients can be very large in number and face challenges of data and network heterogeneity. Examples of clients include smartphones and connected vehicles, which highlights the practical relevance of federated learning. We benchmark three federated learning algorithms and compare their performance against a centralized approach where data resides on the server. The algorithms Federated Averaging (FedAvg), Federated Stochastic Variance Reduced Gradient, and CO-OP are evaluated on the MNIST dataset, using both i.i.d. and non-i.i.d. partitionings of the data. Our results show that FedAvg achieves the highest accuracy among the federated algorithms, regardless of how data was partitioned. Our comparison between FedAvg and centralized learning shows that they are practically equivalent when i.i.d. data is used. However, the centralized approach outperforms FedAvg with non-i.i.d. data."
]
} |
1901.09774 | 2912208857 | Facial attributes are important since they provide a detailed description and determine the visual appearance of human faces. In this paper, we aim at converting a face image to a sketch while simultaneously generating facial attributes. To this end, we propose a novel Attribute-Guided Sketch Generative Adversarial Network (ASGAN) which is an end-to-end framework and contains two pairs of generators and discriminators, one of which is used to generate faces with attributes while the other one is employed for image-to-sketch translation. The two generators form a W-shaped network (W-net) and they are trained jointly with a weight-sharing constraint. Additionally, we also propose two novel discriminators, the residual one focusing on attribute generation and the triplex one helping to generate realistic looking sketches. To validate our model, we have created a new large dataset with 8,804 images, named the Attribute Face Photo & Sketch (AFPS) dataset which is the first dataset containing attributes associated to face sketch images. The experimental results demonstrate that the proposed network (i) generates more photo-realistic faces with sharper facial attributes than baselines and (ii) has good generalization capability on different generative tasks. | Based on CGANs, @cite_46 have developed a generic framework Pix2pix'', which is suitable for different generative tasks. In Pix2pix, one conditional image is adopted as a reference during the training time. The generator in Pix2pix is a U-net, which tries to synthesize a fake image conditioned on the given conditional image in order to fool the discriminator, while the discriminator tries to identify the fake image by comparing it with the corresponding target image. Under these settings, the discriminator takes the pairs of images as input. The U-net is actually an Encoder-Decoder network with skip connection, in which the encoder consists of multiple convolution layers and the decoder consists of multiple deconvolution layers. @cite_46 added skip connections between each layer @math and layer @math which allows feature sharing between the encoder and decoder, where @math is the total number of layers. All channels at layer @math are simply concatenated with those at layer @math by the skip connections. @cite_46 shares a similar goal with us, but it cannot solve the face-to-attributed-sketch translation task since it cannot convert a face image to a sketch conditioned on external facial attributes while our ASGAN is specifically designed to tackle this task. | {
"cite_N": [
"@cite_46"
],
"mid": [
"2772288692"
],
"abstract": [
"Recently, image-to-image translation has been made much progress owing to the success of conditional Generative Adversarial Networks (cGANs). However, it's still very challenging for translation tasks with the requirement of high-level visual information conversion, such as photo-to-caricature translation that requires satire, exaggeration, lifelikeness and artistry. We present an approach for learning to translate faces in the wild from the source photo domain to the target caricature domain with different styles, which can also be used for other high-level image-to-image translation tasks. In order to capture global structure with local statistics while translation, we design a dual pathway model of cGAN with one global discriminator and one patch discriminator. Beyond standard convolution (Conv), we propose a new parallel convolution (ParConv) to construct Parallel Convolutional Neural Networks (ParCNNs) for both global and patch discriminators, which can combine the information from previous layer with the current layer. For generator, we provide three more extra losses in association with adversarial loss to constrain consistency for generated output itself and with the target. Also the style can be controlled by the input style info vector. Experiments on photo-to-caricature translation of faces in the wild show considerable performance gain of our proposed method over state-of-the-art translation methods as well as its potential real applications."
]
} |
1901.09892 | 2914520039 | Neural networks play an increasingly important role in the field of machine learning and are included in many applications in society. Unfortunately, neural networks suffer from adversarial samples generated to attack them. However, most of the generation approaches either assume that the attacker has full knowledge of the neural network model or are limited by the type of attacked model. In this paper, we propose a new approach that generates a black-box attack to neural networks based on the swarm evolutionary algorithm. Benefiting from the improvements in the technology and theoretical characteristics of evolutionary algorithms, our approach has the advantages of effectiveness, black-box attack, generality, and randomness. Our experimental results show that both the MNIST images and the CIFAR-10 images can be perturbed to successful generate a black-box attack with 100 probability on average. In addition, the proposed attack, which is successful on distilled neural networks with almost 100 probability, is resistant to defensive distillation. The experimental results also indicate that the robustness of the artificial intelligence algorithm is related to the complexity of the model and the data set. In addition, we find that the adversarial samples to some extent reproduce the characteristics of the sample data learned by the neural network model. | Some recent research aimed to defend against the attack of adversarial samples and proposed approaches such as defensive distillatione @cite_20 @cite_10 @cite_31 @cite_28 . However, experiment results show that these approaches do not perform well in particular situations due to not being able to defend against adversarial samples of high quality @cite_3 . | {
"cite_N": [
"@cite_28",
"@cite_3",
"@cite_31",
"@cite_10",
"@cite_20"
],
"mid": [
"2964082701",
"2174868984",
"2625220439",
"2799031185",
"2906208681"
],
"abstract": [
"Deep learning algorithms have been shown to perform extremely well on manyclassical machine learning problems. However, recent studies have shown thatdeep learning, like other machine learning techniques, is vulnerable to adversarial samples: inputs crafted to force adeep neural network (DNN) to provide adversary-selected outputs. Such attackscan seriously undermine the security of the system supported by the DNN, sometimes with devastating consequences. For example, autonomous vehicles canbe crashed, illicit or illegal content can bypass content filters, or biometricauthentication systems can be manipulated to allow improper access. In thiswork, we introduce a defensive mechanism called defensive distillationto reduce the effectiveness of adversarial samples on DNNs. We analyticallyinvestigate the generalizability and robustness properties granted by the useof defensive distillation when training DNNs. We also empirically study theeffectiveness of our defense mechanisms on two DNNs placed in adversarialsettings. The study shows that defensive distillation can reduce effectivenessof sample creation from 95 to less than 0.5 on a studied DNN. Such dramaticgains can be explained by the fact that distillation leads gradients used inadversarial sample creation to be reduced by a factor of 1030. We alsofind that distillation increases the average minimum number of features thatneed to be modified to create adversarial samples by about 800 on one of theDNNs we tested.",
"Deep learning algorithms have been shown to perform extremely well on many classical machine learning problems. However, recent studies have shown that deep learning, like other machine learning techniques, is vulnerable to adversarial samples: inputs crafted to force a deep neural network (DNN) to provide adversary-selected outputs. Such attacks can seriously undermine the security of the system supported by the DNN, sometimes with devastating consequences. For example, autonomous vehicles can be crashed, illicit or illegal content can bypass content filters, or biometric authentication systems can be manipulated to allow improper access. In this work, we introduce a defensive mechanism called defensive distillation to reduce the effectiveness of adversarial samples on DNNs. We analytically investigate the generalizability and robustness properties granted by the use of defensive distillation when training DNNs. We also empirically study the effectiveness of our defense mechanisms on two DNNs placed in adversarial settings. The study shows that defensive distillation can reduce effectiveness of sample creation from 95 to less than 0.5 on a studied DNN. Such dramatic gains can be explained by the fact that distillation leads gradients used in adversarial sample creation to be reduced by a factor of 10^30. We also find that distillation increases the average minimum number of features that need to be modified to create adversarial samples by about 800 on one of the DNNs we tested.",
"Ongoing research has proposed several methods to defend neural networks against adversarial examples, many of which researchers have shown to be ineffective. We ask whether a strong defense can be created by combining multiple (possibly weak) defenses. To answer this question, we study three defenses that follow this approach. Two of these are recently proposed defenses that intentionally combine components designed to work well together. A third defense combines three independent defenses. For all the components of these defenses and the combined defenses themselves, we show that an adaptive adversary can create adversarial examples successfully with low distortion. Thus, our work implies that ensemble of weak defenses is not sufficient to provide strong defense against adversarial examples.",
"In recent years, defending adversarial perturbations to natural examples in order to build robust machine learning models trained by deep neural networks (DNNs) has become an emerging research field in the conjunction of deep learning and security. In particular, MagNet consisting of an adversary detector and a data reformer is by far one of the strongest defenses in the black-box oblivious attack setting, where the attacker aims to craft transferable adversarial examples from an undefended DNN model to bypass an unknown defense module deployed on the same DNN model. Under this setting, MagNet can successfully defend a variety of attacks in DNNs, including the high-confidence adversarial examples generated by the Carlini and Wagner's attack based on the @math distortion metric. However, in this paper, under the same attack setting we show that adversarial examples crafted based on the @math distortion metric can easily bypass MagNet and mislead the target DNN image classifiers on MNIST and CIFAR-10. We also provide explanations on why the considered approach can yield adversarial examples with superior attack performance and conduct extensive experiments on variants of MagNet to verify its lack of robustness to @math distortion based attacks. Notably, our results substantially weaken the assumption of effective threat models on MagNet that require knowing the deployed defense technique when attacking DNNs (i.e., the gray-box attack setting).",
"We consider adversarial examples in the black-box decision-based scenario. Here, an attacker has access to the final classification of a model, but not its parameters or softmax outputs. Most attacks for this scenario are based either on transferability, which is unreliable, or random sampling, which is often slow. Focusing on the latter, we propose to improve the efficiency of sampling-based attacks with prior beliefs about the target domain. We identify two such priors, image frequency and surrogate gradients, and discuss how to integrate them into a unified sampling procedure. We then formulate the Biased Boundary Attack, which achieves a drastic speedup over the original Boundary Attack. We demonstrate the effectiveness of our approach against an ImageNet classifier. We also showcase a targeted attack for the Google Cloud Vision API, where we craft convincing examples with just a few hundred queries. Finally, we demonstrate that our approach outperforms the state of the art when facing strong defenses: Our attack scored second place in the targeted attack track of the NeurIPS 2018 Adversarial Vision Challenge."
]
} |
1907.09987 | 2963313981 | Bayesian inference is used extensively to infer and to quantify the uncertainty in a field of interest from a measurement of a related field when the two are linked by a physical model. Despite its many applications, Bayesian inference faces challenges when inferring fields that have discrete representations of large dimension, and or have prior distributions that are difficult to represent mathematically. In this manuscript we consider the use of Generative Adversarial Networks (GANs) in addressing these challenges. A GAN is a type of deep neural network equipped with the ability to learn the distribution implied by multiple samples of a given field. Once trained on these samples, the generator component of a GAN maps the iid components of a low-dimensional latent vector to an approximation of the distribution of the field of interest. In this work we demonstrate how this approximate distribution may be used as a prior in a Bayesian update, and how it addresses the challenges associated with characterizing complex prior distributions and the large dimension of the inferred field. We demonstrate the efficacy of this approach by applying it to the problem of inferring and quantifying uncertainty in the initial temperature field in a heat conduction problem from a noisy measurement of the temperature at later time. | The solution of an inverse problem using sample-based priors has a rich history (see @cite_6 @cite_33 for example). As does the idea of reducing the dimension of the parameter space by mapping it to a lower-dimensional space @cite_17 @cite_27 . However, the use of GANs in these tasks is novel. | {
"cite_N": [
"@cite_27",
"@cite_33",
"@cite_6",
"@cite_17"
],
"mid": [
"2804184144",
"2963105487",
"2074686342",
"2964040595"
],
"abstract": [
"Solving inverse problems continues to be a challenge in a wide array of applications ranging from deblurring, image inpainting, source separation etc. Most existing techniques solve such inverse problems by either explicitly or implicitly finding the inverse of the model. The former class of techniques require explicit knowledge of the measurement process which can be unrealistic, and rely on strong analytical regularizers to constrain the solution space, which often do not generalize well. The latter approaches have had remarkable success in part due to deep learning, but require a large collection of source-observation pairs, which can be prohibitively expensive. In this paper, we propose an unsupervised technique to solve inverse problems with generative adversarial networks (GANs). Using a pre-trained GAN in the space of source signals, we show that one can reliably recover solutions to under determined problems in a blind' fashion, i.e., without knowledge of the measurement process. We solve this by making successive estimates on the model and the solution in an iterative fashion. We show promising results in three challenging applications -- blind source separation, image deblurring, and recovering an image from its edge map, and perform better than several baselines.",
"Generative adversarial networks (GANs) learn a deep generative model that is able to synthesize novel, high-dimensional data samples. New data samples are synthesized by passing latent samples, drawn from a chosen prior distribution, through the generative model. Once trained, the latent space exhibits interesting properties that may be useful for downstream tasks such as classification or retrieval. Unfortunately, GANs do not offer an “inverse model,” a mapping from data space back to latent space, making it difficult to infer a latent representation for a given data sample. In this paper, we introduce a technique, inversion , to project data samples, specifically images, to the latent space using a pretrained GAN. Using our proposed inversion technique, we are able to identify which attributes of a data set a trained GAN is able to model and quantify GAN performance, based on a reconstruction loss. We demonstrate how our proposed inversion technique may be used to quantitatively compare the performance of various GAN models trained on three image data sets. We provide codes for all of our experiments in the website ( https: github.com ToniCreswell InvertingGAN ).",
"We consider a Bayesian approach to nonlinear inverse problems in which the unknown quantity is a spatial or temporal field, endowed with a hierarchical Gaussian process prior. Computational challenges in this construction arise from the need for repeated evaluations of the forward model (e.g., in the context of Markov chain Monte Carlo) and are compounded by high dimensionality of the posterior. We address these challenges by introducing truncated Karhunen-Loeve expansions, based on the prior distribution, to efficiently parameterize the unknown field and to specify a stochastic forward problem whose solution captures that of the deterministic forward model over the support of the prior. We seek a solution of this problem using Galerkin projection on a polynomial chaos basis, and use the solution to construct a reduced-dimensionality surrogate posterior density that is inexpensive to evaluate. We demonstrate the formulation on a transient diffusion equation with prescribed source terms, inferring the spatially-varying diffusivity of the medium from limited and noisy data.",
"Abstract Nonlinear dimensionality reduction embeddings computed from datasets do not provide a mechanism to compute the inverse map. In this paper, we address the problem of computing a stable inverse map to such a general bi-Lipschitz map. Our approach relies on radial basis functions (RBFs) to interpolate the inverse map everywhere on the low-dimensional image of the forward map. We demonstrate that the scale-free cubic RBF kernel performs better than the Gaussian kernel: it does not suffer from ill-conditioning, and does not require the choice of a scale. The proposed construction is shown to be similar to the Nystrom extension of the eigenvectors of the symmetric normalized graph Laplacian matrix. Based on this observation, we provide a new interpretation of the Nystrom extension with suggestions for improvement."
]
} |
1907.09987 | 2963313981 | Bayesian inference is used extensively to infer and to quantify the uncertainty in a field of interest from a measurement of a related field when the two are linked by a physical model. Despite its many applications, Bayesian inference faces challenges when inferring fields that have discrete representations of large dimension, and or have prior distributions that are difficult to represent mathematically. In this manuscript we consider the use of Generative Adversarial Networks (GANs) in addressing these challenges. A GAN is a type of deep neural network equipped with the ability to learn the distribution implied by multiple samples of a given field. Once trained on these samples, the generator component of a GAN maps the iid components of a low-dimensional latent vector to an approximation of the distribution of the field of interest. In this work we demonstrate how this approximate distribution may be used as a prior in a Bayesian update, and how it addresses the challenges associated with characterizing complex prior distributions and the large dimension of the inferred field. We demonstrate the efficacy of this approach by applying it to the problem of inferring and quantifying uncertainty in the initial temperature field in a heat conduction problem from a noisy measurement of the temperature at later time. | Recently, a number of authors have considered the use machine learning-based methods for solving inverse problems. These include the use of convolutional neural networks (CNNs) to solve physics-driven inverse problems @cite_18 @cite_23 @cite_46 , and GANs to solve problems in computer vision @cite_57 @cite_65 @cite_29 @cite_54 @cite_49 @cite_15 @cite_51 @cite_47 . There is also a growing body of work dedicated to using GANs to learn regularizers in solving inverse problems @cite_31 and in compressed sensing @cite_32 @cite_58 @cite_19 @cite_13 @cite_30 . However, these approaches differs from ours in at least two significant ways. First, they solve the inverse problem as an optimization problem and do not rely on Bayesian inference; as a result, regularization is added in an ad-hoc manner, and no attempt is made to quantify the uncertainty in the inferred field. Second, the forward map is assumed to satisfy an extension of the restricted isometry property, which may not be the case for forward maps induced by physics-based operators. | {
"cite_N": [
"@cite_13",
"@cite_30",
"@cite_18",
"@cite_47",
"@cite_31",
"@cite_15",
"@cite_29",
"@cite_54",
"@cite_65",
"@cite_32",
"@cite_57",
"@cite_19",
"@cite_23",
"@cite_49",
"@cite_46",
"@cite_58",
"@cite_51"
],
"mid": [
"2574952845",
"2607406448",
"2804184144",
"2604885021",
"2950788055",
"2166513231",
"2942471690",
"2768812640",
"2949360407",
"2805646839",
"2949257576",
"2887556118",
"2963896595",
"2963226480",
"2762473905",
"2750704532",
"2783032472"
],
"abstract": [
"In this paper, we propose a novel deep convolutional neural network (CNN)-based algorithm for solving ill-posed inverse problems. Regularized iterative algorithms have emerged as the standard approach to ill-posed inverse problems in the past few decades. These methods produce excellent results, but can be challenging to deploy in practice due to factors including the high computational cost of the forward and adjoint operators and the difficulty of hyperparameter selection. The starting point of this paper is the observation that unrolled iterative methods have the form of a CNN (filtering followed by pointwise non-linearity) when the normal operator ( @math , where @math is the adjoint of the forward imaging operator, @math ) of the forward model is a convolution. Based on this observation, we propose using direct inversion followed by a CNN to solve normal-convolutional inverse problems. The direct inversion encapsulates the physical model of the system, but leads to artifacts when the problem is ill posed; the CNN combines multiresolution decomposition and residual learning in order to learn to remove these artifacts while preserving image structure. We demonstrate the performance of the proposed network in sparse-view reconstruction (down to 50 views) on parallel beam X-ray computed tomography in synthetic phantoms as well as in real experimental sinograms. The proposed network outperforms total variation-regularized iterative reconstruction for the more realistic phantoms and requires less than a second to reconstruct a @math image on the GPU.",
"We propose a partially learned approach for the solution of ill-posed inverse problems with not necessarily linear forward operators. The method builds on ideas from classical regularisation theory and recent advances in deep learning to perform learning while making use of prior information about the inverse problem encoded in the forward operator, noise model and a regularising functional. The method results in a gradient-like iterative scheme, where the 'gradient' component is learned using a convolutional network that includes the gradients of the data discrepancy and regulariser as input in each iteration. We present results of such a partially learned gradient scheme on a non-linear tomographic inversion problem with simulated data from both the Sheep-Logan phantom as well as a head CT. The outcome is compared against filtered backprojection and total variation reconstruction and the proposed method provides a 5.4 dB PSNR improvement over the total variation reconstruction while being significantly faster, giving reconstructions of pixel images in about 0.4 s using a single graphics processing unit (GPU).",
"Solving inverse problems continues to be a challenge in a wide array of applications ranging from deblurring, image inpainting, source separation etc. Most existing techniques solve such inverse problems by either explicitly or implicitly finding the inverse of the model. The former class of techniques require explicit knowledge of the measurement process which can be unrealistic, and rely on strong analytical regularizers to constrain the solution space, which often do not generalize well. The latter approaches have had remarkable success in part due to deep learning, but require a large collection of source-observation pairs, which can be prohibitively expensive. In this paper, we propose an unsupervised technique to solve inverse problems with generative adversarial networks (GANs). Using a pre-trained GAN in the space of source signals, we show that one can reliably recover solutions to under determined problems in a blind' fashion, i.e., without knowledge of the measurement process. We solve this by making successive estimates on the model and the solution in an iterative fashion. We show promising results in three challenging applications -- blind source separation, image deblurring, and recovering an image from its edge map, and perform better than several baselines.",
"While deep learning methods have achieved state-of-theart performance in many challenging inverse problems like image inpainting and super-resolution, they invariably involve problem-specific training of the networks. Under this approach, each inverse problem requires its own dedicated network. In scenarios where we need to solve a wide variety of problems, e.g., on a mobile camera, it is inefficient and expensive to use these problem-specific networks. On the other hand, traditional methods using analytic signal priors can be used to solve any linear inverse problem; this often comes with a performance that is worse than learning-based methods. In this work, we provide a middle ground between the two kinds of methods — we propose a general framework to train a single deep neural network that solves arbitrary linear inverse problems. We achieve this by training a network that acts as a quasi-projection operator for the set of natural images and show that any linear inverse problem involving natural images can be solved using iterative methods. We empirically show that the proposed framework demonstrates superior performance over traditional methods using wavelet sparsity prior while achieving performance comparable to specially-trained networks on tasks including compressive sensing and pixel-wise inpainting.",
"While deep learning methods have achieved state-of-the-art performance in many challenging inverse problems like image inpainting and super-resolution, they invariably involve problem-specific training of the networks. Under this approach, different problems require different networks. In scenarios where we need to solve a wide variety of problems, e.g., on a mobile camera, it is inefficient and costly to use these specially-trained networks. On the other hand, traditional methods using signal priors can be used in all linear inverse problems but often have worse performance on challenging tasks. In this work, we provide a middle ground between the two kinds of methods --- we propose a general framework to train a single deep neural network that solves arbitrary linear inverse problems. The proposed network acts as a proximal operator for an optimization algorithm and projects non-image signals onto the set of natural images defined by the decision boundary of a classifier. In our experiments, the proposed framework demonstrates superior performance over traditional methods using a wavelet sparsity prior and achieves comparable performance of specially-trained networks on tasks including compressive sensing and pixel-wise inpainting.",
"Neural networks with their inherent learning ability have been widely applied to solve the robot manipulator inverse kinematics problems. However, there are still two open problems: (1) without knowing inverse kinematic expressions, these solutions have the difficulty of how to collect training sets, and (2) the gradient-based learning algorithms can cause a very slow training process, especially for a complex configuration, or a large set of training data. Unlike these traditional implementations, the proposed metho trains neural network in joint subspace which can be easily calculated with electromagnetism-like method. The kinematics equation and its inverse are one-to-one mapping within the subspace. Thus the constrained training sets can be easily collected by forward kinematics relations. For issue 2, this paper uses a novel learning algorithm called extreme learning machine (ELM) which randomly choose the input weights and analytically determines the output weights of the single hidden layer feedforward neural networks (SLFNs). In theory, this algorithm tends to provide the best generalization performance at extremely fast learning speed. The results show that the proposed approach has not only greatly reduced the computation time but also improved the precision.",
"We introduce a method to infer a variational approximation to the posterior distribution of solutions in computational imaging inverse problems. Machine learning methods applied to computational imaging have proven very successful, but have so far largely focused on retrieving a single optimal solution for a given task. Such retrieval is arguably an incomplete description of the solution space, as in ill-posed inverse problems there may be many similarly likely reconstructions. We minimise an upper bound on the divergence between our approximate distribution and the true intractable posterior, thereby obtaining a probabilistic description of the solution space in imaging inverse problems with empirical prior. We demonstrate the advantage of our technique in quantitative simulations with the CelebA dataset and common image reconstruction tasks. We then apply our method to two of the currently most challenging problems in experimental optics: imaging through highly scattering media and imaging through multi-modal optical fibres. In both settings we report state of the art reconstructions, while providing new capabilities, such as estimation of error-bars and visualisation of multiple likely reconstructions.",
"Generative Adversarial Networks (GAN) have attracted much research attention recently, leading to impressive results for natural image generation. However, to date little success was observed in using GAN generated images for improving classification tasks. Here we attempt to explore, in the context of car license plate recognition, whether it is possible to generate synthetic training data using GAN to improve recognition accuracy. With a carefully-designed pipeline, we show that the answer is affirmative. First, a large-scale image set is generated using the generator of GAN, without manual annotation. Then, these images are fed to a deep convolutional neural network (DCNN) followed by a bidirectional recurrent neural network (BRNN) with long short-term memory (LSTM), which performs the feature learning and sequence labelling. Finally, the pre-trained model is fine-tuned on real images. Our experimental results on a few data sets demonstrate the effectiveness of using GAN images: an improvement of 7.5 over a strong baseline with moderate-sized real data being available. We show that the proposed framework achieves competitive recognition accuracy on challenging test datasets. We also leverage the depthwise separate convolution to construct a lightweight convolutional RNN, which is about half size and 2x faster on CPU. Combining this framework and the proposed pipeline, we make progress in performing accurate recognition on mobile and embedded devices.",
"In many machine learning tasks it is desirable that a model's prediction transforms in an equivariant way under transformations of its input. Convolutional neural networks (CNNs) implement translational equivariance by construction; for other transformations, however, they are compelled to learn the proper mapping. In this work, we develop Steerable Filter CNNs (SFCNNs) which achieve joint equivariance under translations and rotations by design. The proposed architecture employs steerable filters to efficiently compute orientation dependent responses for many orientations without suffering interpolation artifacts from filter rotation. We utilize group convolutions which guarantee an equivariant mapping. In addition, we generalize He's weight initialization scheme to filters which are defined as a linear combination of a system of atomic filters. Numerical experiments show a substantial enhancement of the sample complexity with a growing number of sampled filter orientations and confirm that the network generalizes learned patterns over orientations. The proposed approach achieves state-of-the-art on the rotated MNIST benchmark and on the ISBI 2012 2D EM segmentation challenge.",
"In most computer vision applications, convolutional neural networks (CNNs) operate on dense image data generated by ordinary cameras. Designing CNNs for sparse and irregularly spaced input data is still an open problem with numerous applications in autonomous driving, robotics, and surveillance. To tackle this challenging problem, we introduce an algebraically-constrained convolution layer for CNNs with sparse input and demonstrate its capabilities for the scene depth completion task. We propose novel strategies for determining the confidence from the convolution operation and propagating it to consecutive layers. Furthermore, we propose an objective function that simultaneously minimizes the data error while maximizing the output confidence. Comprehensive experiments are performed on the KITTI depth benchmark and the results clearly demonstrate that the proposed approach achieves superior performance while requiring three times fewer parameters than the state-of-the-art methods. Moreover, our approach produces a continuous pixel-wise confidence map enabling information fusion, state inference, and decision support.",
"The main contribution of this paper is a simple semi-supervised pipeline that only uses the original training set without collecting extra data. It is challenging in 1) how to obtain more training data only from the training set and 2) how to use the newly generated data. In this work, the generative adversarial network (GAN) is used to generate unlabeled samples. We propose the label smoothing regularization for outliers (LSRO). This method assigns a uniform label distribution to the unlabeled images, which regularizes the supervised model and improves the baseline. We verify the proposed method on a practical problem: person re-identification (re-ID). This task aims to retrieve a query person from other cameras. We adopt the deep convolutional generative adversarial network (DCGAN) for sample generation, and a baseline convolutional neural network (CNN) for representation learning. Experiments show that adding the GAN-generated data effectively improves the discriminative ability of learned CNN embeddings. On three large-scale datasets, Market-1501, CUHK03 and DukeMTMC-reID, we obtain +4.37 , +1.6 and +2.46 improvement in rank-1 precision over the baseline CNN, respectively. We additionally apply the proposed method to fine-grained bird recognition and achieve a +0.6 improvement over a strong baseline. The code is available at this https URL",
"Abstract Visual tracking algorithms based on structured output support vector machine (SOSVM) have demonstrated excellent performance. However, sampling methods and optimization strategies of SOSVM undesirably increase the computational overloads, which hinder real-time application of these algorithms. Moreover, due to the lack of high-dimensional features and dense training samples, SOSVM-based algorithms are unstable to deal with various challenging scenarios, such as occlusions and scale variations. Recently, visual tracking algorithms based on discriminative correlation filters (DCF), especially the combination of DCF and features from deep convolutional neural networks (CNN), have been successfully applied to visual tracking, and attains surprisingly good performance on recent benchmarks. The success is mainly attributed to two aspects: the circular correlation properties of DCF and the powerful representation capabilities of CNN features. Nevertheless, compared with SOSVM, DCF-based algorithms are restricted to simple ridge regression which has a weaker discriminative ability. In this paper, a novel circular and structural operator tracker (CSOT) is proposed for high performance visual tracking, it not only possesses the powerful discriminative capability of SOSVM but also efficiently inherits the superior computational efficiency of DCF. Based on the proposed circular and structural operators, a set of primal confidence score maps can be obtained by circular correlating feature maps with their corresponding structural correlation filters. Furthermore, an implicit interpolation is applied to convert the multi-resolution feature maps to the continuous domain and make all primal confidence score maps have the same spatial resolution. Then, we exploit an efficient ensemble post-processor based on relative entropy, which can coalesce primal confidence score maps and create an optimal confidence score map for more accurate localization. The target is localized on the peak of the optimal confidence score map. Besides, we introduce a collaborative optimization strategy to update circular and structural operators by iteratively training structural correlation filters, which significantly reduces computational complexity and improves robustness. Experimental results demonstrate that our approach achieves state-of-the-art performance in mean AUC scores of 71.5 and 69.4 on the OTB2013 and OTB2015 benchmarks respectively, and obtains a third-best expected average overlap (EAO) score of 29.8 on the VOT2017 benchmark.",
"Conventional deep convolutional neural networks (CNNs) apply convolution operators uniformly in space across all feature maps for hundreds of layers - this incurs a high computational cost for real-time applications. For many problems such as object detection and semantic segmentation, we are able to obtain a low-cost computation mask, either from a priori problem knowledge, or from a low-resolution segmentation network. We show that such computation masks can be used to reduce computation in the high-resolution main network. Variants of sparse activation CNNs have previously been explored on small-scale tasks and showed no degradation in terms of object classification accuracy, but often measured gains in terms of theoretical FLOPs without realizing a practical speedup when compared to highly optimized dense convolution implementations. In this work, we leverage the sparsity structure of computation masks and propose a novel tiling-based sparse convolution algorithm. We verified the effectiveness of our sparse CNN on LiDAR-based 3D object detection, and we report significant wall-clock speed-ups compared to dense convolution without noticeable loss of accuracy.",
"Inverse Problems in medical imaging and computer vision are traditionally solved using purely model-based methods. Among those variational regularization models are one of the most popular approaches. We propose a new framework for applying data-driven approaches to inverse problems, using a neural network as a regularization functional. The network learns to discriminate between the distribution of ground truth images and the distribution of unregularized reconstructions. Once trained, the network is applied to the inverse problem by solving the corresponding variational problem. Unlike other data-based approaches for inverse problems, the algorithm can be applied even if only unsupervised training data is available. Experiments demonstrate the potential of the framework for denoising on the BSDS dataset and for computer tomography reconstruction on the LIDC dataset.",
"Optimizing deep neural networks (DNNs) often suffers from the ill-conditioned problem. We observe that the scaling-based weight space symmetry property in rectified nonlinear network will cause this negative effect. Therefore, we propose to constrain the incoming weights of each neuron to be unit-norm, which is formulated as an optimization problem over Oblique manifold. A simple yet efficient method referred to as projection based weight normalization (PBWN) is also developed to solve this problem. PBWN executes standard gradient updates, followed by projecting the updated weight back to Oblique manifold. This proposed method has the property of regularization and collaborates well with the commonly used batch normalization technique. We conduct comprehensive experiments on several widely-used image datasets including CIFAR-10, CIFAR-100, SVHN and ImageNet for supervised learning over the state-of-the-art convolutional neural networks, such as Inception, VGG and residual networks. The results show that our method is able to improve the performance of DNNs with different architectures consistently. We also apply our method to Ladder network for semi-supervised learning on permutation invariant MNIST dataset, and our method outperforms the state-of-the-art methods: we obtain test errors as 2.52 , 1.06 , and 0.91 with only 20, 50, and 100 labeled samples, respectively.",
"Convolutional neural networks (CNNs) have recently been applied to the optical flow estimation problem. As training the CNNs requires sufficiently large ground truth training data, existing approaches resort to synthetic, unrealistic datasets. On the other hand, unsupervised methods are capable of leveraging real-world videos for training where the ground truth flow fields are not available. These methods, however, rely on the fundamental assumptions of brightness constancy and spatial smoothness priors which do not hold near motion boundaries. In this paper, we propose to exploit unlabeled videos for semi-supervised learning of optical flow with a Generative Adversarial Network. Our key insight is that the adversarial loss can capture the structural patterns of flow warp errors without making explicit assumptions. Extensive experiments on benchmark datasets demonstrate that the proposed semi-supervised algorithm performs favorably against purely supervised and semi-supervised learning schemes.",
"Conventional deep convolutional neural networks (CNNs) apply convolution operators uniformly in space across all feature maps for hundreds of layers - this incurs a high computational cost for real time applications. For many problems such as object detection and semantic segmentation, we are able to obtain a low-cost computation mask, either from a priori problem knowledge, or from a low resolution segmentation network. We show that such computation masks can be used to reduce computation in the high resolution main network. Variants of sparse activation CNNs have previously been explored on small scale tasks, and showed no degradation in terms of object classification accuracy, but often measured gains in terms of theoretical FLOPs without realizing a practical speed-up when compared to highly optimized dense convolution implementations. In this work, we leverage the sparsity structure of computation masks and propose a novel tiling-based sparse convolution algorithm. We verified the effectiveness of our sparse CNN on LiDAR based 3D object detection, and we report significant wall-clock speed-ups compared to dense convolution, as well as improved detection accuracy."
]
} |
1907.09987 | 2963313981 | Bayesian inference is used extensively to infer and to quantify the uncertainty in a field of interest from a measurement of a related field when the two are linked by a physical model. Despite its many applications, Bayesian inference faces challenges when inferring fields that have discrete representations of large dimension, and or have prior distributions that are difficult to represent mathematically. In this manuscript we consider the use of Generative Adversarial Networks (GANs) in addressing these challenges. A GAN is a type of deep neural network equipped with the ability to learn the distribution implied by multiple samples of a given field. Once trained on these samples, the generator component of a GAN maps the iid components of a low-dimensional latent vector to an approximation of the distribution of the field of interest. In this work we demonstrate how this approximate distribution may be used as a prior in a Bayesian update, and how it addresses the challenges associated with characterizing complex prior distributions and the large dimension of the inferred field. We demonstrate the efficacy of this approach by applying it to the problem of inferring and quantifying uncertainty in the initial temperature field in a heat conduction problem from a noisy measurement of the temperature at later time. | More recently, the approach described in @cite_36 utilizes GANs in a Bayesian setting; however the GAN is trained to approximate the posterior distribution (and not the prior, as in our case), and training is done in a supervised fashion. That is, paired samples of the measurement @math and the corresponding true solution @math are required. In contrast, our approach is unsupervised, where we require only samples of the true solution @math to train the GAN prior. | {
"cite_N": [
"@cite_36"
],
"mid": [
"2607491080"
],
"abstract": [
"Traditional generative adversarial networks (GAN) and many of its variants are trained by minimizing the KL or JS-divergence loss that measures how close the generated data distribution is from the true data distribution. A recent advance called the WGAN based on Wasserstein distance can improve on the KL and JS-divergence based GANs, and alleviate the gradient vanishing, instability, and mode collapse issues that are common in the GAN training. In this work, we aim at improving on the WGAN by first generalizing its discriminator loss to a margin-based one, which leads to a better discriminator, and in turn a better generator, and then carrying out a progressive training paradigm involving multiple GANs to contribute to the maximum margin ranking loss so that the GAN at later stages will improve upon early stages. We call this method Gang of GANs (GoGAN). We have shown theoretically that the proposed GoGAN can reduce the gap between the true data distribution and the generated data distribution by at least half in an optimally trained WGAN. We have also proposed a new way of measuring GAN quality which is based on image completion tasks. We have evaluated our method on four visual datasets: CelebA, LSUN Bedroom, CIFAR-10, and 50K-SSFF, and have seen both visual and quantitative improvement over baseline WGAN."
]
} |
1907.09798 | 2972934250 | Motivated by the success of encoding multi-scale contextual information for image analysis, we propose our PointAtrousGraph (PAG) - a deep permutation-invariant hierarchical encoder-decoder for efficiently exploiting multi-scale edge features in point clouds. Our PAG is constructed by several novel modules, such as Point Atrous Convolution (PAC), Edge-preserved Pooling (EP) and Edge-preserved Unpooling (EU). Similar with atrous convolution, our PAC can effectively enlarge receptive fields of filters and thus densely learn multi-scale point features. Following the idea of non-overlapping max-pooling operations, we propose our EP to preserve critical edge features during subsampling. Correspondingly, our EU modules gradually recover spatial information for edge features. In addition, we introduce chained skip subsampling upsampling modules that directly propagate edge features to the final stage. Particularly, our proposed auxiliary loss functions can further improve our performance. Experimental results show that our PAG outperform previous state-of-the-art methods on various 3D semantic perception applications. | Deep hierarchical encoder-decoder architectures are widely and successfully used for many image-based tasks, such as human pose estimation @cite_41 @cite_80 , semantic segmentation @cite_66 @cite_3 @cite_11 @cite_60 @cite_48 @cite_73 @cite_64 @cite_33 @cite_46 @cite_72 , optical flow estimation @cite_53 @cite_49 , and object detection @cite_6 @cite_74 @cite_56 . The encoder-decoder architecture, stacked hourglass module, is based on the successive steps of pooling and upsampling, which produces impressive results on human pose estimation @cite_41 . @cite_6 introduced the feature pyramid network for object detection. As for semantic segmentation tasks, U-Net @cite_66 , SegNet @cite_3 and DeconvNet @cite_72 follow the symmetric encoder-decoder architectures, and they refine the segmentation masks by utilizing features in low-level layers. DeepLabv3+ @cite_15 takes advantage of both the encoder-decoder architecture and the atrous convolution modules to effectively change the fields-of-view of filters to capture multi-scale contextual information, which provides new state-of-the-art performance on many semantic segmentation benchmarks. that progressively reduces the feature resolution, enlarges the receptive fields of filters and captures higher semantic information; (2) that gradually recovers the spatial information @cite_15 . | {
"cite_N": [
"@cite_64",
"@cite_33",
"@cite_60",
"@cite_41",
"@cite_48",
"@cite_53",
"@cite_15",
"@cite_3",
"@cite_6",
"@cite_56",
"@cite_72",
"@cite_49",
"@cite_74",
"@cite_80",
"@cite_46",
"@cite_73",
"@cite_66",
"@cite_11"
],
"mid": [
"2963881378",
"1910657905",
"2787091153",
"2964309882",
"1969366022",
"2963727650",
"2587989515",
"2789983685",
"2067912884",
"2952600975",
"2602709638",
"2526041965",
"2768430110",
"2783000019",
"2796438033",
"2401944043",
"360623563",
"2785325870"
],
"abstract": [
"We present a novel and practical deep fully convolutional neural network architecture for semantic pixel-wise segmentation termed SegNet. This core trainable segmentation engine consists of an encoder network, a corresponding decoder network followed by a pixel-wise classification layer. The architecture of the encoder network is topologically identical to the 13 convolutional layers in the VGG16 network [1] . The role of the decoder network is to map the low resolution encoder feature maps to full input resolution feature maps for pixel-wise classification. The novelty of SegNet lies is in the manner in which the decoder upsamples its lower resolution input feature map(s). Specifically, the decoder uses pooling indices computed in the max-pooling step of the corresponding encoder to perform non-linear upsampling. This eliminates the need for learning to upsample. The upsampled maps are sparse and are then convolved with trainable filters to produce dense feature maps. We compare our proposed architecture with the widely adopted FCN [2] and also with the well known DeepLab-LargeFOV [3] , DeconvNet [4] architectures. This comparison reveals the memory versus accuracy trade-off involved in achieving good segmentation performance. SegNet was primarily motivated by scene understanding applications. Hence, it is designed to be efficient both in terms of memory and computational time during inference. It is also significantly smaller in the number of trainable parameters than other competing architectures and can be trained end-to-end using stochastic gradient descent. We also performed a controlled benchmark of SegNet and other architectures on both road scenes and SUN RGB-D indoor scene segmentation tasks. These quantitative assessments show that SegNet provides good performance with competitive inference time and most efficient inference memory-wise as compared to other architectures. We also provide a Caffe implementation of SegNet and a web demo at http: mi.eng.cam.ac.uk projects segnet .",
"We present a novel and practical deep fully convolutional neural network architecture for semantic pixel-wise segmentation termed SegNet. This core trainable segmentation engine consists of an encoder network, a corresponding decoder network followed by a pixel-wise classification layer. The architecture of the encoder network is topologically identical to the 13 convolutional layers in the VGG16 network. The role of the decoder network is to map the low resolution encoder feature maps to full input resolution feature maps for pixel-wise classification. The novelty of SegNet lies is in the manner in which the decoder upsamples its lower resolution input feature map(s). Specifically, the decoder uses pooling indices computed in the max-pooling step of the corresponding encoder to perform non-linear upsampling. This eliminates the need for learning to upsample. The upsampled maps are sparse and are then convolved with trainable filters to produce dense feature maps. We compare our proposed architecture with the widely adopted FCN and also with the well known DeepLab-LargeFOV, DeconvNet architectures. This comparison reveals the memory versus accuracy trade-off involved in achieving good segmentation performance. SegNet was primarily motivated by scene understanding applications. Hence, it is designed to be efficient both in terms of memory and computational time during inference. It is also significantly smaller in the number of trainable parameters than other competing architectures. We also performed a controlled benchmark of SegNet and other architectures on both road scenes and SUN RGB-D indoor scene segmentation tasks. We show that SegNet provides good performance with competitive inference time and more efficient inference memory-wise as compared to other architectures. We also provide a Caffe implementation of SegNet and a web demo at this http URL",
"Spatial pyramid pooling module or encode-decoder structure are used in deep neural networks for semantic segmentation task. The former networks are able to encode multi-scale contextual information by probing the incoming features with filters or pooling operations at multiple rates and multiple effective fields-of-view, while the latter networks can capture sharper object boundaries by gradually recovering the spatial information. In this work, we propose to combine the advantages from both methods. Specifically, our proposed model, DeepLabv3+, extends DeepLabv3 by adding a simple yet effective decoder module to refine the segmentation results especially along object boundaries. We further explore the Xception model and apply the depthwise separable convolution to both Atrous Spatial Pyramid Pooling and decoder modules, resulting in a faster and stronger encoder-decoder network. We demonstrate the effectiveness of the proposed model on PASCAL VOC 2012 and Cityscapes datasets, achieving the test set performance of 89.0 and 82.1 without any post-processing. Our paper is accompanied with a publicly available reference implementation of the proposed models in Tensorflow at this https URL .",
"Spatial pyramid pooling module or encode-decoder structure are used in deep neural networks for semantic segmentation task. The former networks are able to encode multi-scale contextual information by probing the incoming features with filters or pooling operations at multiple rates and multiple effective fields-of-view, while the latter networks can capture sharper object boundaries by gradually recovering the spatial information. In this work, we propose to combine the advantages from both methods. Specifically, our proposed model, DeepLabv3+, extends DeepLabv3 by adding a simple yet effective decoder module to refine the segmentation results especially along object boundaries. We further explore the Xception model and apply the depthwise separable convolution to both Atrous Spatial Pyramid Pooling and decoder modules, resulting in a faster and stronger encoder-decoder network. We demonstrate the effectiveness of the proposed model on PASCAL VOC 2012 and Cityscapes datasets, achieving the test set performance of 89 and 82.1 without any post-processing. Our paper is accompanied with a publicly available reference implementation of the proposed models in Tensorflow at https: github.com tensorflow models tree master research deeplab.",
"In this paper, we address the problems of contour detection, bottom-up grouping, object detection and semantic segmentation on RGB-D data. We focus on the challenging setting of cluttered indoor scenes, and evaluate our approach on the recently introduced NYU-Depth V2 (NYUD2) dataset (, ECCV, 2012). We propose algorithms for object boundary detection and hierarchical segmentation that generalize the @math gPb-ucm approach of (TPAMI, 2011) by making effective use of depth information. We show that our system can label each contour with its type (depth, normal or albedo). We also propose a generic method for long-range amodal completion of surfaces and show its effectiveness in grouping. We train RGB-D object detectors by analyzing and computing histogram of oriented gradients on the depth image and using them with deformable part models (, TPAMI, 2010). We observe that this simple strategy for training object detectors significantly outperforms more complicated models in the literature. We then turn to the problem of semantic segmentation for which we propose an approach that classifies superpixels into the dominant object categories in the NYUD2 dataset. We design generic and class-specific features to encode the appearance and geometry of objects. We also show that additional features computed from RGB-D object detectors and scene classifiers further improves semantic segmentation accuracy. In all of these tasks, we report significant improvements over the state-of-the-art.",
"Recent work has made significant progress in improving spatial resolution for pixelwise labeling with Fully Convolutional Network (FCN) framework by employing Dilated Atrous convolution, utilizing multi-scale features and refining boundaries. In this paper, we explore the impact of global contextual information in semantic segmentation by introducing the Context Encoding Module, which captures the semantic context of scenes and selectively highlights class-dependent featuremaps. The proposed Context Encoding Module significantly improves semantic segmentation results with only marginal extra computation cost over FCN. Our approach has achieved new state-of-the-art results 51.7 mIoU on PASCAL-Context, 85.9 mIoU on PASCAL VOC 2012. Our single model achieves a final score of 0.5567 on ADE20K test set, which surpasses the winning entry of COCO-Place Challenge 2017. In addition, we also explore how the Context Encoding Module can improve the feature representation of relatively shallow networks for the image classification on CIFAR-10 dataset. Our 14 layer network has achieved an error rate of 3.45 , which is comparable with state-of-the-art approaches with over 10A— more layers. The source code for the complete system are publicly available1.",
"In this paper we address the problem of semantic labeling of indoor scenes on RGB-D data. With the availability of RGB-D cameras, it is expected that additional depth measurement will improve the accuracy. Here we investigate a solution how to incorporate complementary depth information into a semantic segmentation framework by making use of convolutional neural networks (CNNs). Recently encoder-decoder type fully convolutional CNN architectures have achieved a great success in the field of semantic segmentation. Motivated by this observation we propose an encoder-decoder type network, where the encoder part is composed of two branches of networks that simultaneously extract features from RGB and depth images and fuse depth features into the RGB feature maps as the network goes deeper. Comprehensive experimental evaluations demonstrate that the proposed fusion-based architecture achieves competitive results with the state-of-the-art methods on the challenging SUN RGB-D benchmark obtaining 76.27 global accuracy, 48.30 average class accuracy and 37.29 average intersection-over-union score.",
"Recent work has made significant progress in improving spatial resolution for pixelwise labeling with Fully Convolutional Network (FCN) framework by employing Dilated Atrous convolution, utilizing multi-scale features and refining boundaries. In this paper, we explore the impact of global contextual information in semantic segmentation by introducing the Context Encoding Module, which captures the semantic context of scenes and selectively highlights class-dependent featuremaps. The proposed Context Encoding Module significantly improves semantic segmentation results with only marginal extra computation cost over FCN. Our approach has achieved new state-of-the-art results 51.7 mIoU on PASCAL-Context, 85.9 mIoU on PASCAL VOC 2012. Our single model achieves a final score of 0.5567 on ADE20K test set, which surpass the winning entry of COCO-Place Challenge in 2017. In addition, we also explore how the Context Encoding Module can improve the feature representation of relatively shallow networks for the image classification on CIFAR-10 dataset. Our 14 layer network has achieved an error rate of 3.45 , which is comparable with state-of-the-art approaches with over 10 times more layers. The source code for the complete system are publicly available.",
"We address the problems of contour detection, bottom-up grouping and semantic segmentation using RGB-D data. We focus on the challenging setting of cluttered indoor scenes, and evaluate our approach on the recently introduced NYU-Depth V2 (NYUD2) dataset [27]. We propose algorithms for object boundary detection and hierarchical segmentation that generalize the gPb-ucm approach of [2] by making effective use of depth information. We show that our system can label each contour with its type (depth, normal or albedo). We also propose a generic method for long-range amodal completion of surfaces and show its effectiveness in grouping. We then turn to the problem of semantic segmentation and propose a simple approach that classifies super pixels into the 40 dominant object categories in NYUD2. We use both generic and class-specific features to encode the appearance and geometry of objects. We also show how our approach can be used for scene classification, and how this contextual information in turn improves object recognition. In all of these tasks, we report significant improvements over the state-of-the-art.",
"In this paper, we propose an encoder-decoder convolutional neural network (CNN) architecture for estimating camera pose (orientation and location) from a single RGB-image. The architecture has a hourglass shape consisting of a chain of convolution and up-convolution layers followed by a regression part. The up-convolution layers are introduced to preserve the fine-grained information of the input image. Following the common practice, we train our model in end-to-end manner utilizing transfer learning from large scale classification data. The experiments demonstrate the performance of the approach on data exhibiting different lighting conditions, reflections, and motion blur. The results indicate a clear improvement over the previous state-of-the-art even when compared to methods that utilize sequence of test frames instead of a single frame.",
"In this paper, we propose an encoder-decoder convolutional neural network (CNN) architecture for estimating camera pose (orientation and location) from a single RGB-image. The architecture has a hourglass shape consisting of a chain of convolution and up-convolution layers followed by a regression part. The up-convolution layers are introduced to preserve the fine-grained information of the input image. Following the common practice, we train our model in end-to-end manner utilizing transfer learning from large scale classification data. The experiments demonstrate the performance of the approach on data exhibiting different lighting conditions, reflections, and motion blur The results indicate a clear improvement over the previous state-of-the-art even when compared to methods that utilize sequence of test frames instead of a single frame.",
"We explore architectures for general pixel-level prediction problems, from low-level edge detection to mid-level surface normal estimation to high-level semantic segmentation. Convolutional predictors, such as the fully-convolutional network (FCN), have achieved remarkable success by exploiting the spatial redundancy of neighboring pixels through convolutional processing. Though computationally efficient, we point out that such approaches are not statistically efficient during learning precisely because spatial redundancy limits the information learned from neighboring pixels. We demonstrate that (1) stratified sampling allows us to add diversity during batch updates and (2) sampled multi-scale features allow us to explore more nonlinear predictors (multiple fully-connected layers followed by ReLU) that improve overall accuracy. Finally, our objective is to show how a architecture can get performance better than (or comparable to) the architectures designed for a particular task. Interestingly, our single architecture produces state-of-the-art results for semantic segmentation on PASCAL-Context, surface normal estimation on NYUDv2 dataset, and edge detection on BSDS without contextual post-processing.",
"All the existing image steganography methods use manually crafted features to hide binary payloads into cover images. This leads to small payload capacity and image distortion. Here we propose a convolutional neural network based encoder-decoder architecture for embedding of images as payload. To this end, we make following three major contributions: (i) we propose a deep learning based generic encoder-decoder architecture for image steganography; (ii) we introduce a new loss function that ensures joint end-to-end training of encoder-decoder networks; (iii) we perform extensive empirical evaluation of proposed architecture on a range of challenging publicly available datasets (MNIST, CIFAR10, PASCAL-VOC12, ImageNet, LFW) and report state-of-the-art payload capacity at high PSNR and SSIM values.",
"In this paper we describe a new mobile architecture, MobileNetV2, that improves the state of the art performance of mobile models on multiple tasks and benchmarks as well as across a spectrum of different model sizes. We also describe efficient ways of applying these mobile models to object detection in a novel framework we call SSDLite. Additionally, we demonstrate how to build mobile semantic segmentation models through a reduced form of DeepLabv3 which we call Mobile DeepLabv3. The MobileNetV2 architecture is based on an inverted residual structure where the input and output of the residual block are thin bottleneck layers opposite to traditional residual models which use expanded representations in the input an MobileNetV2 uses lightweight depthwise convolutions to filter features in the intermediate expansion layer. Additionally, we find that it is important to remove non-linearities in the narrow layers in order to maintain representational power. We demonstrate that this improves performance and provide an intuition that led to this design. Finally, our approach allows decoupling of the input output domains from the expressiveness of the transformation, which provides a convenient framework for further analysis. We measure our performance on Imagenet classification, COCO object detection, VOC image segmentation. We evaluate the trade-offs between accuracy, and number of operations measured by multiply-adds (MAdd), as well as the number of parameters",
"In this paper we describe a new mobile architecture, MobileNetV2, that improves the state of the art performance of mobile models on multiple tasks and benchmarks as well as across a spectrum of different model sizes. We also describe efficient ways of applying these mobile models to object detection in a novel framework we call SSDLite. Additionally, we demonstrate how to build mobile semantic segmentation models through a reduced form of DeepLabv3 which we call Mobile DeepLabv3. The MobileNetV2 architecture is based on an inverted residual structure where the input and output of the residual block are thin bottleneck layers opposite to traditional residual models which use expanded representations in the input an MobileNetV2 uses lightweight depthwise convolutions to filter features in the intermediate expansion layer. Additionally, we find that it is important to remove non-linearities in the narrow layers in order to maintain representational power. We demonstrate that this improves performance and provide an intuition that led to this design. Finally, our approach allows decoupling of the input output domains from the expressiveness of the transformation, which provides a convenient framework for further analysis. We measure our performance on Imagenet classification, COCO object detection, VOC image segmentation. We evaluate the trade-offs between accuracy, and number of operations measured by multiply-adds (MAdd), as well as the number of parameters",
"We propose a structured prediction architecture, which exploits the local generic features extracted by Convolutional Neural Networks and the capacity of Recurrent Neural Networks (RNN) to retrieve distant dependencies. The proposed architecture, called ReSeg, is based on the recently introduced ReNet model for image classification. We modify and extend it to perform the more challenging task of semantic segmentation. Each ReNet layer is composed of four RNN that sweep the image horizontally and vertically in both directions, encoding patches or activations, and providing relevant global information. Moreover, ReNet layers are stacked on top of pre-trained convolutional layers, benefiting from generic local features. Upsampling layers follow ReNet layers to recover the original image resolution in the final predictions. The proposed ReSeg architecture is efficient, flexible and suitable for a variety of semantic segmentation tasks. We evaluate ReSeg on several widely-used semantic segmentation datasets: Weizmann Horse, Oxford Flower, and CamVid; achieving state-of-the-art performance. Results show that ReSeg can act as a suitable architecture for semantic segmentation tasks, and may have further applications in other structured prediction problems. The source code and model hyperparameters are available on this https URL",
"We propose a novel deep architecture, SegNet, for semantic pixel wise image labelling. SegNet has several attractive properties; (i) it only requires forward evaluation of a fully learnt function to obtain smooth label predictions, (ii) with increasing depth, a larger context is considered for pixel labelling which improves accuracy, and (iii) it is easy to visualise the effect of feature activation(s) in the pixel label space at any depth. SegNet is composed of a stack of encoders followed by a corresponding decoder stack which feeds into a soft-max classification layer. The decoders help map low resolution feature maps at the output of the encoder stack to full input image size feature maps. This addresses an important drawback of recent deep learning approaches which have adopted networks designed for object categorization for pixel wise labelling. These methods lack a mechanism to map deep layer feature maps to input dimensions. They resort to ad hoc methods to upsample features, e.g. by replication. This results in noisy predictions and also restricts the number of pooling layers in order to avoid too much upsampling and thus reduces spatial context. SegNet overcomes these problems by learning to map encoder outputs to image pixel labels. We test the performance of SegNet on outdoor RGB scenes from CamVid, KITTI and indoor scenes from the NYU dataset. Our results show that SegNet achieves state-of-the-art performance even without use of additional cues such as depth, video frames or post-processing with CRF models.",
"Over the last years, deep convolutional neural networks (ConvNets) have transformed the field of computer vision thanks to their unparalleled capacity to learn high level semantic image features. However, in order to successfully learn those features, they usually require massive amounts of manually labeled data, which is both expensive and impractical to scale. Therefore, unsupervised semantic feature learning, i.e., learning without requiring manual annotation effort, is of crucial importance in order to successfully harvest the vast amount of visual data that are available today. In our work we propose to learn image features by training ConvNets to recognize the 2d rotation that is applied to the image that it gets as input. We demonstrate both qualitatively and quantitatively that this apparently simple task actually provides a very powerful supervisory signal for semantic feature learning. We exhaustively evaluate our method in various unsupervised feature learning benchmarks and we exhibit in all of them state-of-the-art performance. Specifically, our results on those benchmarks demonstrate dramatic improvements w.r.t. prior state-of-the-art approaches in unsupervised representation learning and thus significantly close the gap with supervised feature learning. For instance, in PASCAL VOC 2007 detection task our unsupervised pre-trained AlexNet model achieves the state-of-the-art (among unsupervised methods) mAP of 54.4 that is only 2.4 points lower from the supervised case. We get similarly striking results when we transfer our unsupervised learned features on various other tasks, such as ImageNet classification, PASCAL classification, PASCAL segmentation, and CIFAR-10 classification. The code and models of our paper will be published on: this https URL ."
]
} |
1907.09798 | 2972934250 | Motivated by the success of encoding multi-scale contextual information for image analysis, we propose our PointAtrousGraph (PAG) - a deep permutation-invariant hierarchical encoder-decoder for efficiently exploiting multi-scale edge features in point clouds. Our PAG is constructed by several novel modules, such as Point Atrous Convolution (PAC), Edge-preserved Pooling (EP) and Edge-preserved Unpooling (EU). Similar with atrous convolution, our PAC can effectively enlarge receptive fields of filters and thus densely learn multi-scale point features. Following the idea of non-overlapping max-pooling operations, we propose our EP to preserve critical edge features during subsampling. Correspondingly, our EU modules gradually recover spatial information for edge features. In addition, we introduce chained skip subsampling upsampling modules that directly propagate edge features to the final stage. Particularly, our proposed auxiliary loss functions can further improve our performance. Experimental results show that our PAG outperform previous state-of-the-art methods on various 3D semantic perception applications. | Unorganized point cloud is a simple and straight-forward representation of 3D structures. @cite_45 . The pioneering work PointNet @cite_45 achieves permutation-invariance by applying symmetric functions. Inspired by PointNet, many following works @cite_65 @cite_82 @cite_59 @cite_21 propose more complicated symmetric operations to exploit local geometrical details in 3D points. Semantic labeling on point cloud is more challenging than classification and object-part segmentation. SPG @cite_0 and SGPN @cite_50 both construct super point graphs to refine their semantic labeling results. RSNet @cite_32 introduces a slice pooling layer, a recurrent neural network layer and a slice unpooling layer, which projects unordered point features onto an ordered sequence of feature vectors. However, unlike many networks for semantic labeling tasks on images, they @cite_0 @cite_76 @cite_70 do not have hierarchical encoder-decoder architectures. | {
"cite_N": [
"@cite_70",
"@cite_21",
"@cite_65",
"@cite_32",
"@cite_0",
"@cite_45",
"@cite_59",
"@cite_50",
"@cite_76",
"@cite_82"
],
"mid": [
"2895472109",
"2963517242",
"2778361827",
"2796426482",
"2963336905",
"2900731076",
"2796040722",
"2963053547",
"2742069755",
"2850910281"
],
"abstract": [
"Semantic segmentation of 3D unstructured point clouds remains an open research problem. Recent works predict semantic labels of 3D points by virtue of neural networks but take limited context knowledge into consideration. In this paper, a novel end-to-end approach for unstructured point cloud semantic segmentation, named 3P-RNN, is proposed to exploit the inherent contextual features. First the efficient pointwise pyramid pooling module is investigated to capture local structures at various densities by taking multi-scale neighborhood into account. Then the two-direction hierarchical recurrent neural networks (RNNs) are utilized to explore long-range spatial dependencies. Each recurrent layer takes as input the local features derived from unrolled cells and sweeps the 3D space along two directions successively to integrate structure knowledge. On challenging indoor and outdoor 3D datasets, the proposed framework demonstrates robust performance superior to state-of-the-arts.",
"Point clouds are an efficient data format for 3D data. However, existing 3D segmentation methods for point clouds either do not model local dependencies [21] or require added computations [14, 23]. This work presents a novel 3D segmentation framework, RSNet1, to efficiently model local structures in point clouds. The key component of the RSNet is a lightweight local dependency module. It is a combination of a novel slice pooling layer, Recurrent Neural Network (RNN) layers, and a slice unpooling layer. The slice pooling layer is designed to project features of unordered points onto an ordered sequence of feature vectors so that traditional end-to-end learning algorithms (RNNs) can be applied. The performance of RSNet is validated by comprehensive experiments on the S3DIS[1], ScanNet[3], and ShapeNet [34] datasets. In its simplest form, RSNets surpass all previous state-of-the-art methods on these benchmarks. And comparisons against previous state-of-the-art methods [21, 23] demonstrate the efficiency of RSNets.",
"Recent deep networks that directly handle points in a point set, e.g., PointNet, have been state-of-the-art for supervised semantic learning tasks on point clouds such as classification and segmentation. In this work, a novel end-to-end deep auto-encoder is proposed to address unsupervised learning challenges on point clouds. On the encoder side, a graph-based enhancement is enforced to promote local structures on top of PointNet. Then, a novel folding-based approach is proposed in the decoder, which folds a 2D grid onto the underlying 3D object surface of a point cloud. The proposed decoder only uses about 7 parameters of a decoder with fully-connected neural networks, yet leads to a more discriminative representation that achieves higher linear SVM classification accuracy than the benchmark. In addition, the proposed decoder structure is shown, in theory, to be a generic architecture that is able to reconstruct an arbitrary point cloud from a 2D grid. Finally, this folding-based decoder is interpretable since the reconstruction could be viewed as a fine granular warping from the 2D grid to the point cloud surface.",
"Recent deep networks that directly handle points in a point set, e.g., PointNet, have been state-of-the-art for supervised learning tasks on point clouds such as classification and segmentation. In this work, a novel end-to-end deep auto-encoder is proposed to address unsupervised learning challenges on point clouds. On the encoder side, a graph-based enhancement is enforced to promote local structures on top of PointNet. Then, a novel folding-based decoder deforms a canonical 2D grid onto the underlying 3D object surface of a point cloud, achieving low reconstruction errors even for objects with delicate structures. The proposed decoder only uses about 7 parameters of a decoder with fully-connected neural networks, yet leads to a more discriminative representation that achieves higher linear SVM classification accuracy than the benchmark. In addition, the proposed decoder structure is shown, in theory, to be a generic architecture that is able to reconstruct an arbitrary point cloud from a 2D grid. Our code is available at http: www.merl.com research license#FoldingNet",
"Semantic parsing of large-scale 3D point clouds is an important research topic in computer vision and remote sensing fields. Most existing approaches utilize hand-crafted features for each modality independently and combine them in a heuristic manner. They often fail to consider the consistency and complementary information among features adequately, which makes them difficult to capture high-level semantic structures. The features learned by most of the current deep learning methods can obtain high-quality image classification results. However, these methods are hard to be applied to recognize 3D point clouds due to unorganized distribution and various point density of data. In this paper, we propose a 3DCNN-DQN-RNN method which fuses the 3D convolutional neural network (CNN), Deep Q-Network (DQN) and Residual recurrent neural network (RNN)for an efficient semantic parsing of large-scale 3D point clouds. In our method, an eye window under control of the 3D CNN and DQN can localize and segment the points of the object's class efficiently. The 3D CNN and Residual RNN further extract robust and discriminative features of the points in the eye window, and thus greatly enhance the parsing accuracy of large-scale point clouds. Our method provides an automatic process that maps the raw data to the classification results. It also integrates object localization, segmentation and classification into one framework. Experimental results demonstrate that the proposed method outperforms the state-of-the-art point cloud classification methods.",
"Unlike images which are represented in regular dense grids, 3D point clouds are irregular and unordered, hence applying convolution on them can be difficult. In this paper, we extend the dynamic filter to a new convolution operation, named PointConv. PointConv can be applied on point clouds to build deep convolutional networks. We treat convolution kernels as nonlinear functions of the local coordinates of 3D points comprised of weight and density functions. With respect to a given point, the weight functions are learned with multi-layer perceptron networks and density functions through kernel density estimation. The most important contribution of this work is a novel reformulation proposed for efficiently computing the weight functions, which allowed us to dramatically scale up the network and significantly improve its performance. The learned convolution kernel can be used to compute translation-invariant and permutation-invariant convolution on any point set in the 3D space. Besides, PointConv can also be used as deconvolution operators to propagate features from a subsampled point cloud back to its original resolution. Experiments on ModelNet40, ShapeNet, and ScanNet show that deep convolutional neural networks built on PointConv are able to achieve state-of-the-art on challenging semantic segmentation benchmarks on 3D point clouds. Besides, our experiments converting CIFAR-10 into a point cloud showed that networks built on PointConv can match the performance of convolutional networks in 2D images of a similar structure.",
"Unlike on images, semantic learning on 3D point clouds using a deep network is challenging due to the naturally unordered data structure. Among existing works, PointNet has achieved promising results by directly learning on point sets. However, it does not take full advantage of a point's local neighborhood that contains fine-grained structural information which turns out to be helpful towards better semantic learning. In this regard, we present two new operations to improve PointNet with a more efficient exploitation of local structures. The first one focuses on local 3D geometric structures. In analogy to a convolution kernel for images, we define a point-set kernel as a set of learnable 3D points that jointly respond to a set of neighboring data points according to their geometric affinities measured by kernel correlation, adapted from a similar technique for point cloud registration. The second one exploits local high-dimensional feature structures by recursive feature aggregation on a nearest-neighbor-graph computed from 3D positions. Experiments show that our network can efficiently capture local information and robustly achieve better performances on major datasets. Our code is available at this http URL",
"Unlike on images, semantic learning on 3D point clouds using a deep network is challenging due to the naturally unordered data structure. Among existing works, PointNet has achieved promising results by directly learning on point sets. However, it does not take full advantage of a point's local neighborhood that contains fine-grained structural information which turns out to be helpful towards better semantic learning. In this regard, we present two new operations to improve PointNet with a more efficient exploitation of local structures. The first one focuses on local 3D geometric structures. In analogy to a convolution kernel for images, we define a point-set kernel as a set of learnable 3D points that jointly respond to a set of neighboring data points according to their geometric affinities measured by kernel correlation, adapted from a similar technique for point cloud registration. The second one exploits local high-dimensional feature structures by recursive feature aggregation on a nearest-neighbor-graph computed from 3D positions. Experiments show that our network can efficiently capture local information and robustly achieve better performances on major datasets. Our code is available at http: www.merl.com research license#KCNet",
"We present an algorithm for registration between a large-scale point cloud and a close-proximity scanned point cloud, providing a localization solution that is fully independent of prior information about the initial positions of the two point cloud coordinate systems. The algorithm, denoted LORAX, selects super-points–local subsets of points–and describes the geometric structure of each with a low-dimensional descriptor. These descriptors are then used to infer potential matching regions for an efficient coarse registration process, followed by a fine-tuning stage. The set of super-points is selected by covering the point clouds with overlapping spheres, and then filtering out those of low-quality or nonsalient regions. The descriptors are computed using state-of-the-art unsupervised machine learning, utilizing the technology of deep neural network based auto-encoders. Abstract This novel framework provides a strong alternative to the common practice of using manually designed key-point descriptors for coarse point cloud registration. Utilizing super-points instead of key-points allows the available geometrical data to be better exploited to find the correct transformation. Encoding local 3D geometric structures using a deep neural network auto-encoder instead of traditional descriptors continues the trend seen in other computer vision applications and indeed leads to superior results. The algorithm is tested on challenging point cloud registration datasets, and its advantages over previous approaches as well as its robustness to density changes, noise, and missing data are shown.",
"We present multiresolution tree-structured networks to process point clouds for 3D shape understanding and generation tasks. Our network represents a 3D shape as a set of locality-preserving 1D ordered list of points at multiple resolutions. This allows efficient feed-forward processing through 1D convolutions, coarse-to-fine analysis through a multi-grid architecture, and it leads to faster convergence and small memory footprint during training. The proposed tree-structured encoders can be used to classify shapes and outperform existing point-based architectures on shape classification benchmarks, while tree-structured decoders can be used for generating point clouds directly and they outperform existing approaches for image-to-shape inference tasks learned using the ShapeNet dataset. Our model also allows unsupervised learning of point-cloud based shapes by using a variational autoencoder, leading to higher-quality generated shapes."
]
} |
1907.09760 | 2963048191 | We present NOLBO, a variational observation model estimation for 3D multi-object from 2D single shot. Previous probabilistic instance-level understandings mainly consider the single-object image, not single shot with multi-object; relations between objects and the entire scene are out of their focus. The objectness of each observation also hardly join their model. Therefore, we propose a method to approximate the Bayesian observation model of scene-level 3D multi-object understanding. By exploiting variational auto-encoder (VAE), we estimate latent variables from the entire scene, which follow tractable distributions and concurrently imply 3D full shape and pose. To perform object-oriented data association and probabilistic simultaneous localization and mapping (SLAM), our observation models can easily be adopted to probabilistic inference by replacing object-oriented features with latent variables. | With the recent advent of neural networks, a number of single object classification and detection methods with high performance have been proposed @cite_31 @cite_36 @cite_39 . Beyond obtaining one feature vector from one image for an object, several multi-object detection techniques from single shot have been developed by introducing new network structures @cite_12 @cite_59 @cite_51 @cite_16 @cite_38 . In particular, some of these methods can be applied to various real-time tasks since the whole detection network is composed of single network pipeline. | {
"cite_N": [
"@cite_38",
"@cite_36",
"@cite_39",
"@cite_59",
"@cite_31",
"@cite_16",
"@cite_51",
"@cite_12"
],
"mid": [
"2130306094",
"2750784772",
"2807333821",
"2335901184",
"2613886079",
"2342242867",
"2410641892",
"2134557905"
],
"abstract": [
"Deep Neural Networks (DNNs) have recently shown outstanding performance on image classification tasks [14]. In this paper we go one step further and address the problem of object detection using DNNs, that is not only classifying but also precisely localizing objects of various classes. We present a simple and yet powerful formulation of object detection as a regression problem to object bounding box masks. We define a multi-scale inference procedure which is able to produce high-resolution object detections at a low cost by a few network applications. State-of-the-art performance of the approach is shown on Pascal VOC.",
"Despite significant accuracy improvement in convolutional neural networks (CNN) based object detectors, they often require prohibitive runtimes to process an image for real-time applications. State-of-the-art models often use very deep networks with a large number of floating point operations. Efforts such as model compression learn compact models with fewer number of parameters, but with much reduced accuracy. In this work, we propose a new framework to learn compact and fast object detection networks with improved accuracy using knowledge distillation [20] and hint learning [34]. Although knowledge distillation has demonstrated excellent improvements for simpler classification setups, the complexity of detection poses new challenges in the form of regression, region proposals and less voluminous labels. We address this through several innovations such as a weighted cross-entropy loss to address class imbalance, a teacher bounded loss to handle the regression component and adaptation layers to better learn from intermediate teacher distributions. We conduct comprehensive empirical evaluation with different distillation configurations over multiple datasets including PASCAL, KITTI, ILSVRC and MS-COCO. Our results show consistent improvement in accuracy-speed trade-offs for modern multi-class detection models.",
"Abstract We present a novel convolutional neural network (CNN) based pipeline which can effectively fuse multi-level information extracted from different intermediate layers generating hybrid convolutional features (HCF) for edge detection. Different from previous methods, the proposed method fuses multi-level information in a feature-map based manner. The produced hybrid convolutional features can be used to perform high-quality edge detection. The edge detector is also computationally efficient, because it detects edges in an image-to-image way without any post-processing. We evaluate the proposed method on three widely used datasets for edge detection including BSDS500, NYUD and Multicue, and also test the method on Pascal VOC’12 dataset for object contour detection. The results show that HCF achieves an improvement in performance over the state-of-the-art methods on all four datasets. On BSDS500 dataset, the efficient version of the proposed approach achieves ODS F-score of 0.804 with a speed of 22 fps and the high-accuracy version achieves ODS F-score of 0.814 with 11 fps .",
"Deep Convolution Neural Networks (CNNs) have shown impressive performance in various vision tasks such as image classification, object detection and semantic segmentation. For object detection, particularly in still images, the performance has been significantly increased last year thanks to powerful deep networks (e.g. GoogleNet) and detection frameworks (e.g. Regions with CNN features (RCNN)). The lately introduced ImageNet [6] task on object detection from video (VID) brings the object detection task into the video domain, in which objects' locations at each frame are required to be annotated with bounding boxes. In this work, we introduce a complete framework for the VID task based on still-image object detection and general object tracking. Their relations and contributions in the VID task are thoroughly studied and evaluated. In addition, a temporal convolution network is proposed to incorporate temporal information to regularize the detection results and shows its effectiveness for the task. Code is available at https: github.com myfavouritekk vdetlib.",
"Since the introduction of deep convolutional neural networks (CNNs), object detection in imagery has witnessed substantial breakthroughs in state-of-the-art performance. The defense community utilizes overhead image sensors that acquire large field-of-view aerial imagery in various bands of the electromagnetic spectrum, which is then exploited for various applications, including the detection and localization of man-made objects. In this work, we utilize a recent state-of-the art object detection algorithm, faster R-CNN, to train a deep CNN for vehicle detection in multimodal imagery. We utilize the vehicle detection in aerial imagery (VEDAI) dataset, which contains overhead imagery that is representative of an ISR setting. Our contribution includes modification of key parameters in the faster R-CNN algorithm for this setting where the objects of interest are spatially small, occupying less than 1:5×10-3 of the total image pixels. Our experiments show that (1) an appropriately trained deep CNN leads to average precision rates above 93 on vehicle detection, and (2) transfer learning between imagery modalities is possible, yielding average precision rates above 90 in the absence of fine-tuning.",
"In Convolutional Neural Network (CNN)-based object detection methods, region proposal becomes a bottleneck when objects exhibit significant scale variation, occlusion or truncation. In addition, these methods mainly focus on 2D object detection and cannot estimate detailed properties of objects. In this paper, we propose subcategory-aware CNNs for object detection. We introduce a novel region proposal network that uses subcategory information to guide the proposal generating process, and a new detection network for joint detection and subcategory classification. By using subcategories related to object pose, we achieve state of-the-art performance on both detection and pose estimation on commonly used benchmarks.",
"Convolutional neural networks (CNNs) have shown great performance as general feature representations for object recognition applications. However, for multi-label images that contain multiple objects from different categories, scales and locations, global CNN features are not optimal. In this paper, we incorporate local information to enhance the feature discriminative power. In particular, we first extract object proposals from each image. With each image treated as a bag and object proposals extracted from it treated as instances, we transform the multi-label recognition problem into a multi-class multi-instance learning problem. Then, in addition to extracting the typical CNN feature representation from each proposal, we propose to make use of ground-truth bounding box annotations (strong labels) to add another level of local information by using nearest-neighbor relationships of local regions to form a multi-view pipeline. The proposed multi-view multiinstance framework utilizes both weak and strong labels effectively, and more importantly it has the generalization ability to even boost the performance of unseen categories by partial strong labels from other categories. Our framework is extensively compared with state-of-the-art handcrafted feature based methods and CNN based methods on two multi-label benchmark datasets. The experimental results validate the discriminative power and the generalization ability of the proposed framework. With strong labels, our framework is able to achieve state-of-the-art results in both datasets.",
"We assess the applicability of several popular learning methods for the problem of recognizing generic visual categories with invariance to pose, lighting, and surrounding clutter. A large dataset comprising stereo image pairs of 50 uniform-colored toys under 36 azimuths, 9 elevations, and 6 lighting conditions was collected (for a total of 194,400 individual images). The objects were 10 instances of 5 generic categories: four-legged animals, human figures, airplanes, trucks, and cars. Five instances of each category were used for training, and the other five for testing. Low-resolution grayscale images of the objects with various amounts of variability and surrounding clutter were used for training and testing. Nearest neighbor methods, support vector machines, and convolutional networks, operating on raw pixels or on PCA-derived features were tested. Test error rates for unseen object instances placed on uniform backgrounds were around 13 for SVM and 7 for convolutional nets. On a segmentation recognition task with highly cluttered images, SVM proved impractical, while convolutional nets yielded 16 7 error. A real-time version of the system was implemented that can detect and classify objects in natural scenes at around 10 frames per second."
]
} |
1907.09760 | 2963048191 | We present NOLBO, a variational observation model estimation for 3D multi-object from 2D single shot. Previous probabilistic instance-level understandings mainly consider the single-object image, not single shot with multi-object; relations between objects and the entire scene are out of their focus. The objectness of each observation also hardly join their model. Therefore, we propose a method to approximate the Bayesian observation model of scene-level 3D multi-object understanding. By exploiting variational auto-encoder (VAE), we estimate latent variables from the entire scene, which follow tractable distributions and concurrently imply 3D full shape and pose. To perform object-oriented data association and probabilistic simultaneous localization and mapping (SLAM), our observation models can easily be adopted to probabilistic inference by replacing object-oriented features with latent variables. | Various studies have also been conducted to understand the instance-level representation from 2D images such as object shape, orientation or bounding box. @cite_4 @cite_2 @cite_55 and @cite_48 estimate the orientation of the object by viewpoint classification with discretized bins. In addition, 3D bounding box regression has been carried out to obtain the object location and orientation @cite_49 @cite_6 @cite_10 @cite_40 . In order to estimate the distinct 3D shape of objects, @cite_20 aligns the prior shape to a single object image through key point matching and estimates its 3D shape and orientation together. @cite_0 estimates the 3D mesh with linear combination of parameterized prior shapes. In @cite_23 @cite_37 @cite_17 @cite_41 , they have actively utilized non-linear regression and latent variables of neural networks for 3D reconstruction from 2D. | {
"cite_N": [
"@cite_37",
"@cite_4",
"@cite_41",
"@cite_48",
"@cite_55",
"@cite_6",
"@cite_0",
"@cite_40",
"@cite_49",
"@cite_2",
"@cite_23",
"@cite_10",
"@cite_20",
"@cite_17"
],
"mid": [
"2560544142",
"2950382845",
"2143255850",
"2074208271",
"2065090505",
"2582329586",
"2055686029",
"2020575505",
"2604236302",
"1946609740",
"2114111978",
"2018299767",
"2963547760",
"2462462929"
],
"abstract": [
"We present a method for 3D object detection and pose estimation from a single image. In contrast to current techniques that only regress the 3D orientation of an object, our method first regresses relatively stable 3D object properties using a deep convolutional neural network and then combines these estimates with geometric constraints provided by a 2D object bounding box to produce a complete 3D bounding box. The first network output estimates the 3D object orientation using a novel hybrid discrete-continuous loss, which significantly outperforms the L2 loss. The second output regresses the 3D object dimensions, which have relatively little variance compared to alternatives and can often be predicted for many object types. These estimates, combined with the geometric constraints on translation imposed by the 2D bounding box, enable us to recover a stable and accurate 3D object pose. We evaluate our method on the challenging KITTI object detection benchmark [2] both on the official metric of 3D orientation estimation and also on the accuracy of the obtained 3D bounding boxes. Although conceptually simple, our method outperforms more complex and computationally expensive approaches that leverage semantic segmentation, instance level segmentation and flat ground priors [4] and sub-category detection [23][24]. Our discrete-continuous loss also produces state of the art results for 3D viewpoint estimation on the Pascal 3D+ dataset[26].",
"We present a method for 3D object detection and pose estimation from a single image. In contrast to current techniques that only regress the 3D orientation of an object, our method first regresses relatively stable 3D object properties using a deep convolutional neural network and then combines these estimates with geometric constraints provided by a 2D object bounding box to produce a complete 3D bounding box. The first network output estimates the 3D object orientation using a novel hybrid discrete-continuous loss, which significantly outperforms the L2 loss. The second output regresses the 3D object dimensions, which have relatively little variance compared to alternatives and can often be predicted for many object types. These estimates, combined with the geometric constraints on translation imposed by the 2D bounding box, enable us to recover a stable and accurate 3D object pose. We evaluate our method on the challenging KITTI object detection benchmark both on the official metric of 3D orientation estimation and also on the accuracy of the obtained 3D bounding boxes. Although conceptually simple, our method outperforms more complex and computationally expensive approaches that leverage semantic segmentation, instance level segmentation and flat ground priors and sub-category detection. Our discrete-continuous loss also produces state of the art results for 3D viewpoint estimation on the Pascal 3D+ dataset.",
"We present a dense reconstruction approach that overcomes the drawbacks of traditional multiview stereo by incorporating semantic information in the form of learned category-level shape priors and object detection. Given training data comprised of 3D scans and images of objects from various viewpoints, we learn a prior comprised of a mean shape and a set of weighted anchor points. The former captures the commonality of shapes across the category, while the latter encodes similarities between instances in the form of appearance and spatial consistency. We propose robust algorithms to match anchor points across instances that enable learning a mean shape for the category, even with large shape variations across instances. We model the shape of an object instance as a warped version of the category mean, along with instance-specific details. Given multiple images of an unseen instance, we collate information from 2D object detectors to align the structure from motion point cloud with the mean shape, which is subsequently warped and refined to approach the actual shape. Extensive experiments demonstrate that our model is general enough to learn semantic priors for different object categories, yet powerful enough to reconstruct individual shapes with large variations. Qualitative and quantitative evaluations show that our framework can produce more accurate reconstructions than alternative state-of-the-art multiview stereo systems.",
"We present a framework for robustly estimating registration between a 3D volume image and a 2D projection image and evaluate its precision and robustness in spine interventions for vertebral localization in the presence of anatomical deformation. The framework employs a normalized gradient information similarity metric and multi-start covariance matrix adaptation evolution strategy optimization with local-restarts, which provided improved robustness against deformation and content mismatch. The parallelized implementation allowed orders-of-magnitude acceleration in computation time and improved the robustness of registration via multi-start global optimization. Experiments involved a cadaver specimen and two CT datasets (supine and prone) and 36 C-arm fluoroscopy images acquired with the specimen in four positions (supine, prone, supine with lordosis, prone with kyphosis), three regions (thoracic, abdominal, and lumbar), and three levels of geometric magnification (1.7, 2.0, 2.4). Registration accuracy was evaluated in terms of projection distance error (PDE) between the estimated and true target points in the projection image, including 14 400 random trials (200 trials on the 72 registration scenarios) with initialization error up to ±200 mm and ±10°. The resulting median PDE was better than 0.1 mm in all cases, depending somewhat on the resolution of input CT and fluoroscopy images. The cadaver experiments illustrated the tradeoff between robustness and computation time, yielding a success rate of 99.993 in vertebral labeling (with 'success' defined as PDE <5 mm) using 1,718 664 ± 96 582 function evaluations computed in 54.0 ± 3.5 s on a mid-range GPU (nVidia, GeForce GTX690). Parameters yielding a faster search (e.g., fewer multi-starts) reduced robustness under conditions of large deformation and poor initialization (99.535 success for the same data registered in 13.1 s), but given good initialization (e.g., ±5 mm, assuming a robust initial run) the same registration could be solved with 99.993 success in 6.3 s. The ability to register CT to fluoroscopy in a manner robust to patient deformation could be valuable in applications such as radiation therapy, interventional radiology, and an assistant to target localization (e.g., vertebral labeling) in image-guided spine surgery.",
"In this work we seek to move away from the traditional paradigm for 2D object recognition whereby objects are identified in the image as 2D bounding boxes. We focus instead on: i) detecting objects; ii) identifying their 3D poses; iii) characterizing the geometrical and topological properties of the objects in terms of their aspect configurations in 3D. We call such characterization an object's aspect layout (see Fig. 1). We propose a new model for solving these problems in a joint fashion from a single image for object categories. Our model is constructed upon a novel framework based on conditional random fields with maximal margin parameter estimation. Extensive experiments are conducted to evaluate our model's performance in determining object pose and layout from images. We achieve superior viewpoint accuracy results on three public datasets and show extensive quantitative analysis to demonstrate the ability of accurately recovering the aspect layout of objects.",
"We propose a template matching method for the detection of 2D image objects that are characterized by orientation patterns. Our method is based on data representations via orientation scores, which are functions on the space of positions and orientations, and which are obtained via a wavelet-type transform. This new representation allows us to detect orientation patterns in an intuitive and direct way, namely via cross-correlations. Additionally, we propose a generalized linear regression framework for the construction of suitable templates using smoothing splines. Here, it is important to recognize a curved geometry on the position-orientation domain, which we identify with the Lie group SE(2): the roto-translation group. Templates are then optimized in a B-spline basis, and smoothness is defined with respect to the curved geometry. We achieve state-of-the-art results on three different applications: detection of the optic nerve head in the retina (99.83 percent success rate on 1,737 images), of the fovea in the retina (99.32 percent success rate on 1,616 images), and of the pupil in regular camera images (95.86 percent on 1,521 images). The high performance is due to inclusion of both intensity and orientation features with effective geometric priors in the template matching. Moreover, our method is fast due to a cross-correlation based matching approach.",
"We propose a probabilistic formulation of joint silhouette extraction and 3D reconstruction given a series of calibrated 2D images. Instead of segmenting each image separately in order to construct a 3D surface consistent with the estimated silhouettes, we compute the most probable 3D shape that gives rise to the observed color information. The probabilistic framework, based on Bayesian inference, enables robust 3D reconstruction by optimally taking into account the contribution of all views. We solve the arising maximum a posteriori shape inference in a globally optimal manner by convex relaxation techniques in a spatially continuous representation. For an interactively provided user input in the form of scribbles specifying foreground and background regions, we build corresponding color distributions as multivariate Gaussians and find a volume occupancy that best fits to this data in a variational sense. Compared to classical methods for silhouette-based multiview reconstruction, the proposed approach does not depend on initialization and enjoys significant resilience to violations of the model assumptions due to background clutter, specular reflections, and camera sensor perturbations. In experiments on several real-world data sets, we show that exploiting a silhouette coherency criterion in a multiview setting allows for dramatic improvements of silhouette quality over independent 2D segmentations without any significant increase of computational efforts. This results in more accurate visual hull estimation, needed by a multitude of image-based modeling approaches. We made use of recent advances in parallel computing with a GPU implementation of the proposed method generating reconstructions on volume grids of more than 20 million voxels in up to 4.41 seconds.",
"Geometric 3D reasoning has received renewed attention recently, in the context of visual scene understanding. The level of geometric detail, however, is typically limited to qualitative or coarse-grained quantitative representations. This is linked to the fact that today's object class detectors are tuned towards robust 2D matching rather than accurate 3D pose estimation, encouraged by 2D bounding box-based benchmarks such as Pascal VOC. In this paper, we therefore revisit ideas from the early days of computer vision, namely, 3D geometric object class representations for recognition. These representations can recover geometrically far more accurate object hypotheses than just 2D bounding boxes, including relative 3D positions of object parts. In combination with recent robust techniques for shape description and inference, our approach outperforms state-of-the-art results in 3D pose estimation, while at the same time improving 2D localization. In a series of experiments, we analyze our approach in detail, and demonstrate novel applications enabled by our geometric object class representation, such as fine-grained categorization of cars according to their 3D geometry and ultra-wide baseline matching.",
"We introduce a novel method for 3D object detection and pose estimation from color images only. We first use segmentation to detect the objects of interest in 2D even in presence of partial occlusions and cluttered background. By contrast with recent patch-based methods, we rely on a “holistic” approach: We apply to the detected objects a Convolutional Neural Network (CNN) trained to predict their 3D poses in the form of 2D projections of the corners of their 3D bounding boxes. This, however, is not sufficient for handling objects from the recent T-LESS dataset: These objects exhibit an axis of rotational symmetry, and the similarity of two images of such an object under two different poses makes training the CNN challenging. We solve this problem by restricting the range of poses used for training, and by introducing a classifier to identify the range of a pose at run-time before estimating it. We also use an optional additional step that refines the predicted poses. We improve the state-of-the-art on the LINEMOD dataset from 73.7 [2] to 89.3 of correctly registered RGB frames. We are also the first to report results on the Occlusion dataset [1 ] using color images only. We obtain 54 of frames passing the Pose 6D criterion on average on several sequences of the T-LESS dataset, compared to the 67 of the state-of-the-art [10] on the same sequences which uses both color and depth. The full approach is also scalable, as a single network can be trained for multiple objects simultaneously.",
"Despite the great progress achieved in recognizing objects as 2D bounding boxes in images, it is still very challenging to detect occluded objects and estimate the 3D properties of multiple objects from a single image. In this paper, we propose a novel object representation, 3D Voxel Pattern (3DVP), that jointly encodes the key properties of objects including appearance, 3D shape, viewpoint, occlusion and truncation. We discover 3DVPs in a data-driven way, and train a bank of specialized detectors for a dictionary of 3DVPs. The 3DVP detectors are capable of detecting objects with specific visibility patterns and transferring the meta-data from the 3DVPs to the detected objects, such as 2D segmentation mask, 3D pose as well as occlusion or truncation boundaries. The transferred meta-data allows us to infer the occlusion relationship among objects, which in turn provides improved object recognition results. Experiments are conducted on the KITTI detection benchmark [17] and the outdoor-scene dataset [41]. We improve state-of-the-art results on car detection and pose estimation with notable margins (6 in difficult data of KITTI). We also verify the ability of our method in accurately segmenting objects from the background and localizing them in 3D.",
"Geometric 3D reasoning at the level of objects has received renewed attention recently in the context of visual scene understanding. The level of geometric detail, however, is typically limited to qualitative representations or coarse boxes. This is linked to the fact that today's object class detectors are tuned toward robust 2D matching rather than accurate 3D geometry, encouraged by bounding-box-based benchmarks such as Pascal VOC. In this paper, we revisit ideas from the early days of computer vision, namely, detailed, 3D geometric object class representations for recognition. These representations can recover geometrically far more accurate object hypotheses than just bounding boxes, including continuous estimates of object pose and 3D wireframes with relative 3D positions of object parts. In combination with robust techniques for shape description and inference, we outperform state-of-the-art results in monocular 3D pose estimation. In a series of experiments, we analyze our approach in detail and demonstrate novel applications enabled by such an object class representation, such as fine-grained categorization of cars and bicycles, according to their 3D geometry, and ultrawide baseline matching.",
"In this paper we consider the problem of recovering the free space of an indoor scene from its single image. We show that exploiting the box like geometric structure of furniture and constraints provided by the scene, allows us to recover the extent of major furniture objects in 3D. Our “boxy” detector localizes box shaped objects oriented parallel to the scene across different scales and object types, and thus blocks out the occupied space in the scene. To localize the objects more accurately in 3D we introduce a set of specially designed features that capture the floor contact points of the objects. Image based metrics are not very indicative of performance in 3D. We make the first attempt to evaluate single view based occupancy estimates for 3D errors and propose several task driven performance measures towards it. On our dataset of 592 indoor images marked with full 3D geometry of the scene, we show that: (a) our detector works well using image based metrics; (b) our refinement method produces significant improvements in localization in 3D; and (c) if one evaluates using 3D metrics, our method offers major improvements over other single view based scene geometry estimation methods.",
"From a single image, humans are able to perceive the full 3D shape of an object by exploiting learned shape priors from everyday life. Contemporary single-image 3D reconstruction algorithms aim to solve this task in a similar fashion, but often end up with priors that are highly biased by training classes. Here we present an algorithm, Generalizable Reconstruction (GenRe), designed to capture more generic, class-agnostic shape priors. We achieve this with an inference network and training procedure that combine 2.5D representations of visible surfaces (depth and silhouette), spherical shape representations of both visible and non-visible surfaces, and 3D voxel-based representations, in a principled manner that exploits the causal structure of how 3D shapes give rise to 2D images. Experiments demonstrate that GenRe performs well on single-view shape reconstruction, and generalizes to diverse novel objects from categories not seen during training.",
"In this paper, we propose a non-local structured prior for volumetric multi-view 3D reconstruction. Towards this goal, we present a novel Markov random field model based on ray potentials in which assumptions about large 3D surface patches such as planarity or Manhattan world constraints can be efficiently encoded as probabilistic priors. We further derive an inference algorithm that reasons jointly about voxels, pixels and image segments, and estimates marginal distributions of appearance, occupancy, depth, normals and planarity. Key to tractable inference is a novel hybrid representation that spans both voxel and pixel space and that integrates non-local information from 2D image segmentations in a principled way. We compare our non-local prior to commonly employed local smoothness assumptions and a variety of state-of-the-art volumetric reconstruction baselines on challenging outdoor scenes with textureless and reflective surfaces. Our experiments indicate that regularizing over larger distances has the potential to resolve ambiguities where local regularizers fail."
]
} |
1907.09760 | 2963048191 | We present NOLBO, a variational observation model estimation for 3D multi-object from 2D single shot. Previous probabilistic instance-level understandings mainly consider the single-object image, not single shot with multi-object; relations between objects and the entire scene are out of their focus. The objectness of each observation also hardly join their model. Therefore, we propose a method to approximate the Bayesian observation model of scene-level 3D multi-object understanding. By exploiting variational auto-encoder (VAE), we estimate latent variables from the entire scene, which follow tractable distributions and concurrently imply 3D full shape and pose. To perform object-oriented data association and probabilistic simultaneous localization and mapping (SLAM), our observation models can easily be adopted to probabilistic inference by replacing object-oriented features with latent variables. | Through multi-object detection and instance-level understanding altogether, learning the disentangled representation of multi-object becomes achievable. @cite_6 exploits the yolov2 structure @cite_51 to estimate the 3D bounding box and center of the multi-object and obtains the orientations. In @cite_48 , they estimate the 3D shape rendering and orientation under faster R-CNN structure @cite_12 . They obtain the shape rendering via weighted sum of the parameterized prior shape with PCL. Orientations are estimated by classifying the bins which indicate the discretized object pose. Similarly, in @cite_15 , they design the object observation factor to perform data association for pose SLAM. RoIs for multi-object are obtained by @cite_51 . | {
"cite_N": [
"@cite_48",
"@cite_6",
"@cite_15",
"@cite_51",
"@cite_12"
],
"mid": [
"2560544142",
"2950382845",
"2143255850",
"2604236302",
"2128942651"
],
"abstract": [
"We present a method for 3D object detection and pose estimation from a single image. In contrast to current techniques that only regress the 3D orientation of an object, our method first regresses relatively stable 3D object properties using a deep convolutional neural network and then combines these estimates with geometric constraints provided by a 2D object bounding box to produce a complete 3D bounding box. The first network output estimates the 3D object orientation using a novel hybrid discrete-continuous loss, which significantly outperforms the L2 loss. The second output regresses the 3D object dimensions, which have relatively little variance compared to alternatives and can often be predicted for many object types. These estimates, combined with the geometric constraints on translation imposed by the 2D bounding box, enable us to recover a stable and accurate 3D object pose. We evaluate our method on the challenging KITTI object detection benchmark [2] both on the official metric of 3D orientation estimation and also on the accuracy of the obtained 3D bounding boxes. Although conceptually simple, our method outperforms more complex and computationally expensive approaches that leverage semantic segmentation, instance level segmentation and flat ground priors [4] and sub-category detection [23][24]. Our discrete-continuous loss also produces state of the art results for 3D viewpoint estimation on the Pascal 3D+ dataset[26].",
"We present a method for 3D object detection and pose estimation from a single image. In contrast to current techniques that only regress the 3D orientation of an object, our method first regresses relatively stable 3D object properties using a deep convolutional neural network and then combines these estimates with geometric constraints provided by a 2D object bounding box to produce a complete 3D bounding box. The first network output estimates the 3D object orientation using a novel hybrid discrete-continuous loss, which significantly outperforms the L2 loss. The second output regresses the 3D object dimensions, which have relatively little variance compared to alternatives and can often be predicted for many object types. These estimates, combined with the geometric constraints on translation imposed by the 2D bounding box, enable us to recover a stable and accurate 3D object pose. We evaluate our method on the challenging KITTI object detection benchmark both on the official metric of 3D orientation estimation and also on the accuracy of the obtained 3D bounding boxes. Although conceptually simple, our method outperforms more complex and computationally expensive approaches that leverage semantic segmentation, instance level segmentation and flat ground priors and sub-category detection. Our discrete-continuous loss also produces state of the art results for 3D viewpoint estimation on the Pascal 3D+ dataset.",
"We present a dense reconstruction approach that overcomes the drawbacks of traditional multiview stereo by incorporating semantic information in the form of learned category-level shape priors and object detection. Given training data comprised of 3D scans and images of objects from various viewpoints, we learn a prior comprised of a mean shape and a set of weighted anchor points. The former captures the commonality of shapes across the category, while the latter encodes similarities between instances in the form of appearance and spatial consistency. We propose robust algorithms to match anchor points across instances that enable learning a mean shape for the category, even with large shape variations across instances. We model the shape of an object instance as a warped version of the category mean, along with instance-specific details. Given multiple images of an unseen instance, we collate information from 2D object detectors to align the structure from motion point cloud with the mean shape, which is subsequently warped and refined to approach the actual shape. Extensive experiments demonstrate that our model is general enough to learn semantic priors for different object categories, yet powerful enough to reconstruct individual shapes with large variations. Qualitative and quantitative evaluations show that our framework can produce more accurate reconstructions than alternative state-of-the-art multiview stereo systems.",
"We introduce a novel method for 3D object detection and pose estimation from color images only. We first use segmentation to detect the objects of interest in 2D even in presence of partial occlusions and cluttered background. By contrast with recent patch-based methods, we rely on a “holistic” approach: We apply to the detected objects a Convolutional Neural Network (CNN) trained to predict their 3D poses in the form of 2D projections of the corners of their 3D bounding boxes. This, however, is not sufficient for handling objects from the recent T-LESS dataset: These objects exhibit an axis of rotational symmetry, and the similarity of two images of such an object under two different poses makes training the CNN challenging. We solve this problem by restricting the range of poses used for training, and by introducing a classifier to identify the range of a pose at run-time before estimating it. We also use an optional additional step that refines the predicted poses. We improve the state-of-the-art on the LINEMOD dataset from 73.7 [2] to 89.3 of correctly registered RGB frames. We are also the first to report results on the Occlusion dataset [1 ] using color images only. We obtain 54 of frames passing the Pose 6D criterion on average on several sequences of the T-LESS dataset, compared to the 67 of the state-of-the-art [10] on the same sequences which uses both color and depth. The full approach is also scalable, as a single network can be trained for multiple objects simultaneously.",
"We propose an approach to overcome the two main challenges of 3D multiview object detection and localization: The variation of object features due to changes in the viewpoint and the variation in the size and aspect ratio of the object. Our approach proceeds in three steps. Given an initial bounding box of fixed size, we first refine its aspect ratio and size. We can then predict the viewing angle, under the hypothesis that the bounding box actually contains an object instance. Finally, a classifier tuned to this particular viewpoint checks the existence of an instance. As a result, we can find the object instances and estimate their poses, without having to search over all window sizes and potential orientations. We train and evaluate our method on a new object database specifically tailored for this task, containing real-world objects imaged over a wide range of smoothly varying viewpoints and significant lighting changes. We show that the successive estimations of the bounding box and the viewpoint lead to better localization results."
]
} |
1907.09760 | 2963048191 | We present NOLBO, a variational observation model estimation for 3D multi-object from 2D single shot. Previous probabilistic instance-level understandings mainly consider the single-object image, not single shot with multi-object; relations between objects and the entire scene are out of their focus. The objectness of each observation also hardly join their model. Therefore, we propose a method to approximate the Bayesian observation model of scene-level 3D multi-object understanding. By exploiting variational auto-encoder (VAE), we estimate latent variables from the entire scene, which follow tractable distributions and concurrently imply 3D full shape and pose. To perform object-oriented data association and probabilistic simultaneous localization and mapping (SLAM), our observation models can easily be adopted to probabilistic inference by replacing object-oriented features with latent variables. | These studies are efficient because they mainly concern direct and accurate estimation of the object characteristics through network modeling; on the other hand, the probabilistic observation models are relatively less considered. Although they exploit the neural network for nonlinear regression, approximating the intractable distribution is rarely concerned. Therefore, Bayesian inference with obtained features are challenging; for example, data association for SLAM is considered only in front-end and additional algorithms are necessary to perform loop closing and place recognition @cite_45 @cite_15 . | {
"cite_N": [
"@cite_15",
"@cite_45"
],
"mid": [
"2967631872",
"2889482274"
],
"abstract": [
"We present a Bayesian object observation model for complete probabilistic semantic SLAM. Recent studies on object detection and feature extraction have become important for scene understanding and 3D mapping. However, 3D shape of the object is too complex to formulate the probabilistic observation model; therefore, performing the Bayesian inference of the object-oriented features as well as their pose is less considered. Besides, when the robot equipped with an RGB mono camera only observes the projected single view of an object, a significant amount of the 3D shape information is abandoned. Due to these limitations, semantic SLAM and viewpoint-independent loop closure using volumetric 3D object shape is challenging. In order to enable the complete formulation of probabilistic semantic SLAM, we approximate the observation model of a 3D object with a tractable distribution. We also estimate the variational likelihood from the 2D image of the object to exploit its observed single view. In order to evaluate the proposed method, we perform pose and feature estimation, and demonstrate that the automatic loop closure works seamlessly without additional loop detector in various environments.",
"This paper presents a feature encoding method of complex 3D objects for high-level semantic features. Recent approaches to object recognition methods become important for semantic simultaneous localization and mapping (SLAM). However, there is a lack of consideration of the probabilistic observation model for 3D objects, as the shape of a 3D object basically follows a complex probability distribution. Furthermore, since the mobile robot equipped with a range sensor observes only a single view, much information of the object shape is discarded. These limitations are the major obstacles to semantic SLAM and view-independent loop closure using 3D object shapes as features. In order to enable the numerical analysis for the Bayesian inference, we approximate the true observation model of 3D objects to tractable distributions. Since the observation likelihood can be obtained from the generative model, we formulate the true generative model for 3D object with the Bayesian networks. To capture these complex distributions, we apply a variational auto-encoder. To analyze the approximated distributions and encoded features, we perform classification with maximum likelihood estimation and shape retrieval."
]
} |
1907.09760 | 2963048191 | We present NOLBO, a variational observation model estimation for 3D multi-object from 2D single shot. Previous probabilistic instance-level understandings mainly consider the single-object image, not single shot with multi-object; relations between objects and the entire scene are out of their focus. The objectness of each observation also hardly join their model. Therefore, we propose a method to approximate the Bayesian observation model of scene-level 3D multi-object understanding. By exploiting variational auto-encoder (VAE), we estimate latent variables from the entire scene, which follow tractable distributions and concurrently imply 3D full shape and pose. To perform object-oriented data association and probabilistic simultaneous localization and mapping (SLAM), our observation models can easily be adopted to probabilistic inference by replacing object-oriented features with latent variables. | To handle the intractable target distribution, latent variables can be adopted @cite_46 @cite_18 @cite_34 @cite_13 . In order to understand and utilize the latent space, @cite_58 @cite_9 have studied the relations between latent variables and object visualization by using VAE @cite_34 . However, it is still challenging to apply the proposed method to probabilistic model approximation, as it mainly concentrates on the interpretable graphic codes. To approximate the observation probability, entropy and variational likelihood is exploited in the field of the active vision @cite_22 @cite_14 @cite_7 . Using VAE, @cite_60 @cite_8 have proposed methods to approximate the observation model of 3D objects for Bayesian inference. Based on the ELBO which approximates the observation model, they have shown that how the probabilistic SLAM with data association can be performed with expectation-maximization (EM) algorithm. | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_22",
"@cite_7",
"@cite_60",
"@cite_8",
"@cite_9",
"@cite_34",
"@cite_46",
"@cite_58",
"@cite_13"
],
"mid": [
"2967631872",
"2620364083",
"2913002991",
"2889482274",
"2133027741",
"1777241655",
"2951741291",
"2787512446",
"2963085847",
"2620025707",
"2753738274"
],
"abstract": [
"We present a Bayesian object observation model for complete probabilistic semantic SLAM. Recent studies on object detection and feature extraction have become important for scene understanding and 3D mapping. However, 3D shape of the object is too complex to formulate the probabilistic observation model; therefore, performing the Bayesian inference of the object-oriented features as well as their pose is less considered. Besides, when the robot equipped with an RGB mono camera only observes the projected single view of an object, a significant amount of the 3D shape information is abandoned. Due to these limitations, semantic SLAM and viewpoint-independent loop closure using volumetric 3D object shape is challenging. In order to enable the complete formulation of probabilistic semantic SLAM, we approximate the observation model of a 3D object with a tractable distribution. We also estimate the variational likelihood from the 2D image of the object to exploit its observed single view. In order to evaluate the proposed method, we perform pose and feature estimation, and demonstrate that the automatic loop closure works seamlessly without additional loop detector in various environments.",
"We would like to learn a representation of the data which decomposes an observation into factors of variation which we can independently control. Specifically, we want to use minimal supervision to learn a latent representation that reflects the semantics behind a specific grouping of the data, where within a group the samples share a common factor of variation. For example, consider a collection of face images grouped by identity. We wish to anchor the semantics of the grouping into a relevant and disentangled representation that we can easily exploit. However, existing deep probabilistic models often assume that the observations are independent and identically distributed. We present the Multi-Level Variational Autoencoder (ML-VAE), a new deep probabilistic model for learning a disentangled representation of a set of grouped observations. The ML-VAE separates the latent representation into semantically meaningful parts by working both at the group level and the observation level, while retaining efficient test-time inference. Quantitative and qualitative evaluations show that the ML-VAE model (i) learns a semantically meaningful disentanglement of grouped data, (ii) enables manipulation of the latent representation, and (iii) generalises to unseen groups.",
"With the introduction of the variational autoencoder (VAE), probabilistic latent variable models have received renewed attention as powerful generative models. However, their performance in terms of test likelihood and quality of generated samples has been surpassed by autoregressive models without stochastic units. Furthermore, flow-based models have recently been shown to be an attractive alternative that scales well to high-dimensional data. In this paper we close the performance gap by constructing VAE models that can effectively utilize a deep hierarchy of stochastic variables and model complex covariance structures. We introduce the Bidirectional-Inference Variational Autoencoder (BIVA), characterized by a skip-connected generative model and an inference network formed by a bidirectional stochastic inference path. We show that BIVA reaches state-of-the-art test likelihoods, generates sharp and coherent natural images, and uses the hierarchy of latent variables to capture different aspects of the data distribution. We observe that BIVA, in contrast to recent results, can be used for anomaly detection. We attribute this to the hierarchy of latent variables which is able to extract high-level semantic features. Finally, we extend BIVA to semi-supervised classification tasks and show that it performs comparably to state-of-the-art results by generative adversarial networks.",
"This paper presents a feature encoding method of complex 3D objects for high-level semantic features. Recent approaches to object recognition methods become important for semantic simultaneous localization and mapping (SLAM). However, there is a lack of consideration of the probabilistic observation model for 3D objects, as the shape of a 3D object basically follows a complex probability distribution. Furthermore, since the mobile robot equipped with a range sensor observes only a single view, much information of the object shape is discarded. These limitations are the major obstacles to semantic SLAM and view-independent loop closure using 3D object shapes as features. In order to enable the numerical analysis for the Bayesian inference, we approximate the true observation model of 3D objects to tractable distributions. Since the observation likelihood can be obtained from the generative model, we formulate the true generative model for 3D object with the Bayesian networks. To capture these complex distributions, we apply a variational auto-encoder. To analyze the approximated distributions and encoded features, we perform classification with maximum likelihood estimation and shape retrieval.",
"These lecture notes highlight the mathematical and computational structure relating to the formulation of, and development of algorithms for, the Bayesian approach to inverse problems in differential equations. This approach is fundamental in the quantification of uncertainty within applications in volving the blending of mathematical models with data. The finite dimensional situation is described first, along with some motivational examples. Then the development of probability measures on separable Banach space is undertaken, using a random series over an infinite set of functions to construct draws; these probability measures are used as priors in the Bayesian approach to inverse problems. Regularity of draws from the priors is studied in the natural Sobolev or Besov spaces implied by the choice of functions in the random series construction, and the Kolmogorov continuity theorem is used to extend regularity considerations to the space of Holder continuous functions. Bayes’ theorem is de rived in this prior setting, and here interpreted as finding conditions under which the posterior is absolutely continuous with respect to the prior, and determining a formula for the Radon-Nikodym derivative in terms of the likelihood of the data. Having established the form of the posterior, we then describe various properties common to it in the infinite dimensional setting. These properties include well-posedness, approximation theory, and the existence of maximum a posteriori estimators. We then describe measure-preserving dynamics, again on the infinite dimensional space, including Markov chain-Monte C arlo and sequential Monte Carlo methods, and measure-preserving reversible stochastic differential equations. By formulating the theory and algorithms on the underlying infinite dimensional space, we obtain a framework suitable for rigorous analysis of the accuracy of reconstructions, of computational complexity, as well as naturally constructing algorithms which perform well under mesh refinement, since they are inherently well-defined in infinite dimensions.",
"We propose a probabilistic model to infer supervised latent variables in the Hamming space from observed data. Our model allows simultaneous inference of the number of binary latent variables, and their values. The latent variables preserve neighbourhood structure of the data in a sense that objects in the same semantic concept have similar latent values, and objects in different concepts have dissimilar latent values. We formulate the supervised infinite latent variable problem based on an intuitive principle of pulling objects together if they are of the same type, and pushing them apart if they are not. We then combine this principle with a flexible Indian Buffet Process prior on the latent variables. We show that the inferred supervised latent variables can be directly used to perform a nearest neighbour search for the purpose of retrieval. We introduce a new application of dynamically extending hash codes, and show how to effectively couple the structure of the hash codes with continuously growing structure of the neighbourhood preserving infinite latent feature space.",
"In this paper we present a fully Bayesian latent variable model which exploits conditional nonlinear(in)-dependence structures to learn an efficient latent representation. The latent space is factorized to represent shared and private information from multiple views of the data. In contrast to previous approaches, we introduce a relaxation to the discrete segmentation and allow for a \"softly\" shared latent space. Further, Bayesian techniques allow us to automatically estimate the dimensionality of the latent spaces. The model is capable of capturing structure underlying extremely high dimensional spaces. This is illustrated by modelling unprocessed images with tenths of thousands of pixels. This also allows us to directly generate novel images from the trained model by sampling from the discovered latent spaces. We also demonstrate the model by prediction of human pose in an ambiguous setting. Our Bayesian framework allows us to perform disambiguation in a principled manner by including latent space priors which incorporate the dynamic nature of the data.",
"We extend variational autoencoders (VAEs) to collaborative filtering for implicit feedback. This non-linear probabilistic model enables us to go beyond the limited modeling capacity of linear factor models which still largely dominate collaborative filtering research.We introduce a generative model with multinomial likelihood and use Bayesian inference for parameter estimation. Despite widespread use in language modeling and economics, the multinomial likelihood receives less attention in the recommender systems literature. We introduce a different regularization parameter for the learning objective, which proves to be crucial for achieving competitive performance. Remarkably, there is an efficient way to tune the parameter using annealing. The resulting model and learning algorithm has information-theoretic connections to maximum entropy discrimination and the information bottleneck principle. Empirically, we show that the proposed approach significantly outperforms several state-of-the-art baselines, including two recently-proposed neural network approaches, on several real-world datasets. We also provide extended experiments comparing the multinomial likelihood with other commonly used likelihood functions in the latent factor collaborative filtering literature and show favorable results. Finally, we identify the pros and cons of employing a principled Bayesian inference approach and characterize settings where it provides the most significant improvements.",
"We extend variational autoencoders (VAEs) to collaborative filtering for implicit feedback. This non-linear probabilistic model enables us to go beyond the limited modeling capacity of linear factor models which still largely dominate collaborative filtering research. We introduce a generative model with multinomial likelihood and use Bayesian inference to learn this powerful generative model. Despite widespread use in language modeling and economics, the multinomial likelihood receives less attention in the recommender systems literature. We introduce a different regularization parameter for the learning objective, which proves to be crucial for achieving competitive performance. Remarkably, there is an efficient way to tune the parameter using annealing. The resulting model and learning algorithm has information-theoretic connections to maximum entropy discrimination and the information bottleneck principle. Empirically, we show that the proposed approach significantly outperforms several state-of-the-art baselines, including two recently-proposed neural network approaches, on several real-world datasets. We also provide extended experiments comparing the multinomial likelihood with other commonly used likelihood functions in the latent factor collaborative filtering literature and show favorable results. Finally, we identify the pros and cons of employing a principled Bayesian inference approach and characterize settings where it provides the most significant improvements.",
"We study unsupervised generative modeling in terms of the optimal transport (OT) problem between true (but unknown) data distribution @math and the latent variable model distribution @math . We show that the OT problem can be equivalently written in terms of probabilistic encoders, which are constrained to match the posterior and prior distributions over the latent space. When relaxed, this constrained optimization problem leads to a penalized optimal transport (POT) objective, which can be efficiently minimized using stochastic gradient descent by sampling from @math and @math . We show that POT for the 2-Wasserstein distance coincides with the objective heuristically employed in adversarial auto-encoders (AAE) (, 2016), which provides the first theoretical justification for AAEs known to the authors. We also compare POT to other popular techniques like variational auto-encoders (VAE) (Kingma and Welling, 2014). Our theoretical results include (a) a better understanding of the commonly observed blurriness of images generated by VAEs, and (b) establishing duality between Wasserstein GAN (Arjovsky and Bottou, 2017) and POT for the 1-Wasserstein distance.",
"Learning an interpretable factorised representation of the independent data generative factors of the world without supervision is an important precursor for the development of artificial intelligence that is able to learn and reason in the same way that humans do. We introduce beta-VAE, a new state-of-the-art framework for automated discovery of interpretable factorised latent representations from raw image data in a completely unsupervised manner. Our approach is a modification of the variational autoencoder (VAE) framework. We introduce an adjustable hyperparameter beta that balances latent channel capacity and independence constraints with reconstruction accuracy. We demonstrate that beta-VAE with appropriately tuned beta > 1 qualitatively outperforms VAE (beta = 1), as well as state of the art unsupervised (InfoGAN) and semi-supervised (DC-IGN) approaches to disentangled factor learning on a variety of datasets (celebA, faces and chairs). Furthermore, we devise a protocol to quantitatively compare the degree of disentanglement learnt by different models, and show that our approach also significantly outperforms all baselines quantitatively. Unlike InfoGAN, beta-VAE is stable to train, makes few assumptions about the data and relies on tuning a single hyperparameter, which can be directly optimised through a hyper parameter search using weakly labelled data or through heuristic visual inspection for purely unsupervised data."
]
} |
1907.09825 | 2964205377 | Efficient behavior and trajectory planning is one of the major challenges for automated driving. Especially intersection scenarios are very demanding due to their complexity arising from the variety of maneuver possibilities and other traffic participants. A key challenge is to generate behaviors which optimize the comfort and progress of the ego vehicle but at the same time are not too aggressive towards other traffic participants. In order to maintain real time capability for courteous behavior and trajectory planning, an efficient formulation of the optimal control problem and corresponding solving algorithms are required. Consequently, a novel planning framework is presented which considers comfort and progress as well as the courtesy of actions in a graph-based behavior planning module. Utilizing the low level trajectory generation, the behavior result can be further optimized for driving comfort while satisfying constraints over the whole planning horizon. According experiments show the practicability and real time capability of the framework. | A method for producing courteous behavior is to use a game-theoretic interaction model @cite_15 . The problem is modeled in a way that all agents try to optimize their own behavior. By predicting and considering the reaction to the ego trajectory, courteous behavior can be generated. It is shown that this courtesy leads to better imitation of human behavior @cite_15 . | {
"cite_N": [
"@cite_15"
],
"mid": [
"2132339352"
],
"abstract": [
"Modeling the purposeful behavior of imperfect agents from a small number of observations is a challenging task. When restricted to the single-agent decision-theoretic setting, inverse optimal control techniques assume that observed behavior is an approximately optimal solution to an unknown decision problem. These techniques learn a utility function that explains the example behavior and can then be used to accurately predict or imitate future behavior in similar observed or unobserved situations. In this work, we consider similar tasks in competitive and cooperative multi-agent domains. Here, unlike single-agent settings, a player cannot myopically maximize its reward; it must speculate on how the other agents may act to influence the game's outcome. Employing the game-theoretic notion of regret and the principle of maximum entropy, we introduce a technique for predicting and generalizing behavior."
]
} |
1907.09945 | 2963509006 | Affective computing is a field of great interest in many computer vision applications, including video surveillance, behaviour analysis, and human-robot interaction. Most of the existing literature has addressed this field by analysing different sets of face features. However, in the last decade, several studies have shown how body movements can play a key role even in emotion recognition. The majority of these experiments on the body are performed by trained actors whose aim is to simulate emotional reactions. These unnatural expressions differ from the more challenging genuine emotions, thus invalidating the obtained results. In this paper, a solution for basic non-acted emotion recognition based on 3D skeleton and Deep Neural Networks (DNNs) is provided. The proposed work introduces three majors contributions. First, unlike the current state-of-the-art in non-acted body affect recognition, where only static or global body features are considered, in this work also temporal local movements performed by subjects in each frame are examined. Second, an original set of global and time-dependent features for body movement description is provided. Third, to the best of out knowledge, this is the first attempt to use deep learning methods for non-acted body affect recognition. Due to the novelty of the topic, only the UCLIC dataset is currently considered the benchmark for comparative tests. On the latter, the proposed method outperforms all the competitors. | The most accurate affect recognition systems are based on electroencephalography @cite_27 . Although these systems achieve excellent results, they are limited by the use of dedicated sensors that require a controlled environment. The computer vision applications, on the other hand, are more suitable to work in real and uncontrolled scenarios, thanks to the use of more versatile sensors, such as RGB, RGB-D, or thermal cameras. Most of the vision based methods involve emotion recognition by the analysis of facial expressions @cite_18 @cite_15 @cite_14 . This is due to the presence of a great deal of labelled data in the state-of-the-art, organized in datasets, such as in @cite_4 . Although the face is one of the most discriminative ways to identify people's emotions, it is not always possible to capture facial expressions in large and crowded environments. This aspect motivated researchers to try other solutions, including poses and body movements. | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_4",
"@cite_27",
"@cite_15"
],
"mid": [
"2546815878",
"2011723974",
"2745497104",
"2081835714",
"2748543394"
],
"abstract": [
"Affective computing endows computers with the ability to observe and understand, thus generates a variety of emotional characteristics. Facial expression has continuous signal which is direct expression of emotion. Affective computing via facial expressions has aroused great interest in intelligent interactive and pattern recognition recently. RGB camera is sensitive to surroundings, especially to illumination conditions. 2D images are not robust to human faces which are 3D objects. In this paper, we propose a method that uses noisy depth data produced by the low resolution sensor for robust face recognition. We capture color and depth information by Kinect sensor, extract Facial Feature Points (FFPs) feature vector by Face Tracking SDK, and recognize facial expression by Random Forest (RF) algorithm. This method enables facial expression recognition implementation real-time in intelligent interactive.",
"For facial expression recognition systems to be applicable in the real world, they need to be able to detect and track a previously unseen person's face and its facial movements accurately in realistic environments. A highly plausible solution involves performing a “dense” form of alignment, where 60-70 fiducial facial points are tracked with high accuracy. The problem is that, in practice, this type of dense alignment had so far been impossible to achieve in a generic sense, mainly due to poor reliability and robustness. Instead, many expression detection methods have opted for a “coarse” form of face alignment, followed by an application of a biologically inspired appearance descriptor such as the histogram of oriented gradients or Gabor magnitudes. Encouragingly, recent advances to a number of dense alignment algorithms have demonstrated both high reliability and accuracy for unseen subjects [e.g., constrained local models (CLMs)]. This begs the question: Aside from countering against illumination variation, what do these appearance descriptors do that standard pixel representations do not? In this paper, we show that, when close to perfect alignment is obtained, there is no real benefit in employing these different appearance-based representations (under consistent illumination conditions). In fact, when misalignment does occur, we show that these appearance descriptors do work well by encoding robustness to alignment error. For this work, we compared two popular methods for dense alignment-subject-dependent active appearance models versus subject-independent CLMs-on the task of action-unit detection. These comparisons were conducted through a battery of experiments across various publicly available data sets (i.e., CK+, Pain, M3, and GEMEP-FERA). We also report our performance in the recent 2011 Facial Expression Recognition and Analysis Challenge for the subject-independent task.",
"Automated affective computing in the wild setting is a challenging problem in computer vision. Existing annotated databases of facial expressions in the wild are small and mostly cover discrete emotions (aka the categorical model). There are very limited annotated facial databases for affective computing in the continuous dimensional model (e.g., valence and arousal). To meet this need, we collected, annotated, and prepared for public distribution a new database of facial emotions in the wild (called AffectNet). AffectNet contains more than 1,000,000 facial images from the Internet by querying three major search engines using 1,250 emotion related keywords in six different languages. About half of the retrieved images were manually annotated for the presence of seven discrete facial expressions and the intensity of valence and arousal. AffectNet is by far the largest database of facial expression, valence, and arousal in the wild enabling research in automated facial expression recognition in two different emotion models. Two baseline deep neural networks are used to classify images in the categorical model and predict the intensity of valence and arousal. Various evaluation metrics show that our deep neural network baselines can perform better than conventional machine learning methods and off-the-shelf facial expression recognition systems.",
"In this paper we present the techniques used for the University of Montreal's team submissions to the 2013 Emotion Recognition in the Wild Challenge. The challenge is to classify the emotions expressed by the primary human subject in short video clips extracted from feature length movies. This involves the analysis of video clips of acted scenes lasting approximately one-two seconds, including the audio track which may contain human voices as well as background music. Our approach combines multiple deep neural networks for different data modalities, including: (1) a deep convolutional neural network for the analysis of facial expressions within video frames; (2) a deep belief net to capture audio information; (3) a deep autoencoder to model the spatio-temporal information produced by the human actions depicted within the entire scene; and (4) a shallow network architecture focused on extracted features of the mouth of the primary human subject in the scene. We discuss each of these techniques, their performance characteristics and different strategies to aggregate their predictions. Our best single model was a convolutional neural network trained to predict emotions from static frames using two large data sets, the Toronto Face Database and our own set of faces images harvested from Google image search, followed by a per frame aggregation strategy that used the challenge training data. This yielded a test set accuracy of 35.58 . Using our best strategy for aggregating our top performing models into a single predictor we were able to produce an accuracy of 41.03 on the challenge test set. These compare favorably to the challenge baseline test set accuracy of 27.56 .",
"AimEmotion recognition based on facial expression is an important field in affective computing. Current emotion recognition systems may suffer from two shortcomings: translation in facial image may deteriorate the recognition performance, and the classifier is not robust. MethodTo solve above two problems, our team proposed a novel intelligent emotion recognition system. Our method used stationary wavelet entropy to extract features, and employed a single hidden layer feedforward neural network as the classifier. To prevent the training of the classifier fall into local optimum points, we introduced the Jaya algorithm. ResultsThe simulation results over a 20-subject 700-image dataset showed our algorithm reached an overall accuracy of 96.800.14 . ConclusionThis proposed approach performs better than five state-of-the-art approaches in terms of overall accuracy. Besides, the db4 wavelet performs the best among other whole db wavelet family. The 4-level wavelet decomposition is superior to other levels. In the future, we shall test other advanced features and training algorithms."
]
} |
1907.09883 | 2963801192 | All public blockchains are secured by a proof of opportunity cost among block producers. For example, the security offered by proof-of-work (PoW) systems, like Bitcoin, is due to spent computation; it is work precisely because it cannot be performed for free. In general, more resources provably lost in producing blocks yields more security for the blockchain. When two blockchains share the same mechanism for providing opportunity cost, as is the case when they share the same PoW algorithm, the two chains compete for resources from block producers. Indeed, if there exists a liquid market between resource types, then theoretically all blockchains will compete for resources. In this paper, we show that there exists a resource allocation equilibrium between any two blockchains, which is essentially driven by the fiat value of reward that each chain offers in return for providing security. We go on to prove that this equilibrium is singular and always achieved provided that block producers behave in a greedy, but cautious fashion. The opposite is true when they are overly greedy: resource allocation oscillates in extremes between the two chains. We show that these results hold both in practice and in a block generation simulation. Finally, we demonstrate several applications of this theory including a trustless price-ratio oracle, increased security for blockchains whose coins have lower fiat value, and a quantification of cost to allocating resources away from the equilibrium. | Also closely related is the work of @cite_6 who apply the theory of Potential Games @cite_23 to the problem of miner hash rate allocation across multiple blockchains. They prove that, regardless of individual hash rate and coinbase rewards for each of the blockchains, hash rate allocation will converge to a pure equilibrium provided that miners follow . The model assumes minimal rationality on behalf of the players, i.e., that they follow an arbitrary better response step improving their individual payoffs.'' @cite_6 do not identify a specific equilibrium point, nor do they specify what the should be. But their work anticipates some of the theoretical results we present in . Furthermore, they show that the equilibrium point can be changed by changing a blockchain's coinbase reward, a property that is emergent from the properties of the equilibrium and one that we exploit to increase security in . @cite_16 reached similar conclusions as @cite_6 using a slightly different game theoretical model of hash rate allocation across cryptocurrencies and mining pools. | {
"cite_N": [
"@cite_16",
"@cite_23",
"@cite_6"
],
"mid": [
"2809512071",
"2911474632",
"2556732087"
],
"abstract": [
"Abrupt changes in the miner hash rate applied to a proof-of-work (PoW) blockchain can adversely affect user experience and security. Because different PoW blockchains often share hashing algorithms, miners face a complex choice in deciding how to allocate their hash power among chains. We present an economic model that leverages Modern Portfolio Theory to predict a miner’s allocation over time using price data and inferred risk tolerance. The model matches actual allocations with mean absolute error within 20 for four out of the top five miners active on both Bitcoin (BTC) and Bitcoin Cash (BCH) blockchains. A model of aggregate allocation across those four miners shows excellent agreement in magnitude with the actual aggregate as well a correlation coefficient of 0.649. The accuracy of the aggregate allocation model is also sufficient to explain major historical changes in inter-block time (IBT) for BCH. Because estimates of miner risk are not time-dependent and our model is otherwise price-driven, we are able to use it to anticipate the effect of a major price shock on hash allocation and IBT in the BCH blockchain. Using a Monte Carlo simulation, we show that, despite mitigation by the new difficulty adjustment algorithm, a price drop of 50 could increase the IBT by 50 for at least a day, with a peak delay of 100 .",
"We model the competition over several blockchains characterizing multiple cryptocurrencies as a non-cooperative game. Then, we specialize our results to two instances of the general game, showing properties of the Nash equilibrium. In particular, leveraging results about congestion games, we establish the existence of pure Nash equilibria and provide efficient algorithms for finding such equilibria.",
"We study a setting where a set of players simultaneously invest in a shared resource. The resource has a probability of failure and a return on investment, both of which are functions of the total investment by all players. We use a simple reference dependent preference model to capture players with heterogeneous risk attitudes (risk seeking, risk neutral and risk averse). We show the existence and uniqueness of a pure strategy Nash equilibrium in this setting and examine the effect of different risk attitudes on players' strategies in the presence of uncertainty. In particular, we show that at the equilibrium, risk averse players are pushed out of the resource by risk seeking players. We compare the failure probabilities in the decentralized (game-theoretic) and centralized settings, and show that our proposed game belongs to the class of best response potential games, for which there are simple dynamics that allow all players to converge to the equilibrium."
]
} |
1907.09883 | 2963801192 | All public blockchains are secured by a proof of opportunity cost among block producers. For example, the security offered by proof-of-work (PoW) systems, like Bitcoin, is due to spent computation; it is work precisely because it cannot be performed for free. In general, more resources provably lost in producing blocks yields more security for the blockchain. When two blockchains share the same mechanism for providing opportunity cost, as is the case when they share the same PoW algorithm, the two chains compete for resources from block producers. Indeed, if there exists a liquid market between resource types, then theoretically all blockchains will compete for resources. In this paper, we show that there exists a resource allocation equilibrium between any two blockchains, which is essentially driven by the fiat value of reward that each chain offers in return for providing security. We go on to prove that this equilibrium is singular and always achieved provided that block producers behave in a greedy, but cautious fashion. The opposite is true when they are overly greedy: resource allocation oscillates in extremes between the two chains. We show that these results hold both in practice and in a block generation simulation. Finally, we demonstrate several applications of this theory including a trustless price-ratio oracle, increased security for blockchains whose coins have lower fiat value, and a quantification of cost to allocating resources away from the equilibrium. | Several authors have sought to determine the optimal hash rate allocation between blockchains for miners or mining pools. @cite_7 argue that miners allocate their hash rate between multiple blockchains so as to minimize the risk associated with fluctuations in coin price. @cite_25 make a similar argument except that their measure of risk is volatility in the payout rate between mining pools. @cite_1 extend this model to mining across blockchains with different PoW algorithms. All of the above approaches are complimentary to the present work, which seeks only to explain miner behavior. In fact, miner-specific behavioral choices help to explain why the aggregate hash rate allocation does not fully allocate to one chain over another (see for details). | {
"cite_N": [
"@cite_1",
"@cite_25",
"@cite_7"
],
"mid": [
"2809512071",
"2789931105",
"2944490618"
],
"abstract": [
"Abrupt changes in the miner hash rate applied to a proof-of-work (PoW) blockchain can adversely affect user experience and security. Because different PoW blockchains often share hashing algorithms, miners face a complex choice in deciding how to allocate their hash power among chains. We present an economic model that leverages Modern Portfolio Theory to predict a miner’s allocation over time using price data and inferred risk tolerance. The model matches actual allocations with mean absolute error within 20 for four out of the top five miners active on both Bitcoin (BTC) and Bitcoin Cash (BCH) blockchains. A model of aggregate allocation across those four miners shows excellent agreement in magnitude with the actual aggregate as well a correlation coefficient of 0.649. The accuracy of the aggregate allocation model is also sufficient to explain major historical changes in inter-block time (IBT) for BCH. Because estimates of miner risk are not time-dependent and our model is otherwise price-driven, we are able to use it to anticipate the effect of a major price shock on hash allocation and IBT in the BCH blockchain. Using a Monte Carlo simulation, we show that, despite mitigation by the new difficulty adjustment algorithm, a price drop of 50 could increase the IBT by 50 for at least a day, with a peak delay of 100 .",
"The rise of centralized mining pools for risk sharing does not necessarily undermine the decentralization required for public blockchains. However, mining pools as a financial innovation significantly escalates the arms race among competing miners and thus increases the energy consumption of proof-of-work-based blockchains. Each individual miner's cross-pool diversification and endogenous fees charged by pools generally sustain decentralization --- larger pools better internalize their externality on global hash rates, charge higher fees, attract disproportionately fewer miners, and thus grow slower. Empirical evidence from Bitcoin mining supports our model predictions, and the economic insights apply to many other blockchain protocols, as well as mainstream industries with similar characteristics.",
"Mining is a central operation of all proof-of-work (PoW) based cryptocurrencies. The vast majority of miners today participate in \"mining pools\" instead of \"solo mining\" in order to lower risk and achieve a more steady income. However, this rise of participation in mining pools negatively affects the decentralization levels of most cryptocurrencies. In this work, we look into mining pools from the point of view of a miner: We present an analytical model and implement a computational tool that allows miners to optimally distribute their computational power over multiple pools and PoW cryptocurrencies (i.e. build a mining portfolio), taking into account their risk aversion levels. Our tool allows miners to maximize their risk-adjusted earnings by diversifying across multiple mining pools which enhances PoW decentralization. Finally, we run an experiment in Bitcoin historical data and demonstrate that a miner diversifying over multiple pools, as instructed by our model tool, receives a higher overall Sharpe ratio (i.e. average excess reward over its standard deviation volatility)."
]
} |
1907.09883 | 2963801192 | All public blockchains are secured by a proof of opportunity cost among block producers. For example, the security offered by proof-of-work (PoW) systems, like Bitcoin, is due to spent computation; it is work precisely because it cannot be performed for free. In general, more resources provably lost in producing blocks yields more security for the blockchain. When two blockchains share the same mechanism for providing opportunity cost, as is the case when they share the same PoW algorithm, the two chains compete for resources from block producers. Indeed, if there exists a liquid market between resource types, then theoretically all blockchains will compete for resources. In this paper, we show that there exists a resource allocation equilibrium between any two blockchains, which is essentially driven by the fiat value of reward that each chain offers in return for providing security. We go on to prove that this equilibrium is singular and always achieved provided that block producers behave in a greedy, but cautious fashion. The opposite is true when they are overly greedy: resource allocation oscillates in extremes between the two chains. We show that these results hold both in practice and in a block generation simulation. Finally, we demonstrate several applications of this theory including a trustless price-ratio oracle, increased security for blockchains whose coins have lower fiat value, and a quantification of cost to allocating resources away from the equilibrium. | @cite_22 devised a Markov Decision Process (MDP) for discovering optimal selfish mining @cite_15 strategies. @cite_12 expanded the model to incorporate adjustable network parameters and include analysis of doublespend attacks. @cite_21 extend the MDP of @cite_12 to model mining difficulty adjustment. The biggest differences between these approaches and the present work is that the former analyze optimal behavior in single blockchains while the present work attempts to explain behavior across multiple blockchains. | {
"cite_N": [
"@cite_15",
"@cite_21",
"@cite_22",
"@cite_12"
],
"mid": [
"2911793117",
"2964000194",
"1491322982",
"2151620419"
],
"abstract": [
"Consider a Markov decision process (MDP) that admits a set of state-action features, which can linearly express the process's probabilistic transition model. We propose a parametric Q-learning algorithm that finds an approximate-optimal policy using a sample size proportional to the feature dimension @math and invariant with respect to the size of the state space. To further improve its sample efficiency, we exploit the monotonicity property and intrinsic noise structure of the Bellman operator, provided the existence of anchor state-actions that imply implicit non-negativity in the feature space. We augment the algorithm using techniques of variance reduction, monotonicity preservation, and confidence bounds. It is proved to find a policy which is @math -optimal from any initial state with high probability using @math sample transitions for arbitrarily large-scale MDP with a discount factor @math . A matching information-theoretical lower bound is proved, confirming the sample optimality of the proposed method with respect to all parameters (up to polylog factors).",
"We consider the problem of learning an unknown Markov Decision Process (MDP) that is weakly communicating in the infinite horizon setting. We propose a Thompson Sampling-based reinforcement learning algorithm with dynamic episodes (TSDE). At the beginning of each episode, the algorithm generates a sample from the posterior distribution over the unknown model parameters. It then follows the optimal stationary policy for the sampled model for the rest of the episode. The duration of each episode is dynamically determined by two stopping criteria. The first stopping criterion controls the growth rate of episode length. The second stopping criterion happens when the number of visits to any state-action pair is doubled. We establish @math bounds on expected regret under a Bayesian setting, where @math and @math are the sizes of the state and action spaces, @math is time, and @math is the bound of the span. This regret bound matches the best available bound for weakly communicating MDPs. Numerical results show it to perform better than existing algorithms for infinite horizon MDPs.",
"We consider Markov decision processes (MDPs) with multiple discounted reward objectives. Such MDPs occur in design problems where one wishes to simultaneously optimize several criteria, for example, latency and power. The possible trade-offs between the different objectives are characterized by the Pareto curve. We show that every Pareto-optimal point can be achieved by a memoryless strategy; however, unlike in the single-objective case, the memoryless strategy may require randomization. Moreover, we show that the Pareto curve can be approximated in polynomial time in the size of the MDP. Additionally, we study the problem if a given value vector is realizable by any strategy, and show that it can be decided in polynomial time; but the question whether it is realizable by a deterministic memoryless strategy is NP-complete. These results provide efficient algorithms for design exploration in MDP models with multiple objectives.",
"This article addresses reinforcement learning problems based on factored Markov decision processes MDPs in which the agent must choose among a set of candidate abstractions, each build up from a different combination of state components. We present and evaluate a new approach that can perform effective abstraction selection that is more resource-efficient and or more general than existing approaches. The core of the approach is to make selection of an abstraction part of the learning agent's decision-making process by augmenting the agent's action space with internal actions that select the abstraction it uses. We prove that under certain conditions this approach results in a derived MDP whose solution yields both the optimal abstraction for the original MDP and the optimal policy under that abstraction. We examine our approach in three domains of increasing complexity: contextual bandit problems, episodic MDPs, and general MDPs with context-specific structure. © 2013 Wiley Periodicals, Inc."
]
} |
1907.09786 | 2963372556 | We propose a novel single-step training strategy that allows convolutional encoder-decoder networks that use skip connections, to complete partially observed data by means of hallucination. This strategy is demonstrated for the task of completing 2-D road layouts as well as 3-D vehicle shapes. As input, it takes data from a partially observed domain, for which no ground truth is available, and data from an unpaired prior knowledge domain and trains the network in an end-to-end manner. Our single-step training strategy is compared against two state-of-the-art baselines, one using a two-step auto-encoder training strategy and one using an adversarial strategy. Our novel strategy achieves an improvement up to +12.2 F-measure on the Cityscapes dataset. The learned network intrinsically generalizes better than the baselines on unseen datasets, which is demonstrated by an improvement up to +23.8 F-measure on the unseen KITTI dataset. Moreover, our approach outperforms the baselines using the same backbone network on the 3-D shape completion benchmark by a margin of 0.006 Hamming distance. | Closely related to our task of hallucinating the road layout is image inpainting. In recent years, deep convolutional neural networks (CNNs) enable the possibility of image inpainting with large missing areas, as CNNs can extract abstract semantic information from the observable context. The Context Encoder (CE) @cite_12 network is proposed to inpaint the image with large rectangle areas missing at the image center by applying reconstruction and adversarial loss @cite_13 in training. CE-like networks @cite_21 @cite_8 @cite_5 are proposed with additional discriminative networks applied on locally missing regions, or on the entire image in a patch-wise manner, which are able to perform inpainting with regions missing at arbitrary position. Given a trained generative network, Yeh al @cite_20 do inpainting by finding the embedding vector that minimizes the reconstruction loss by applying back-propagation to the input embedding vector. The main difference between inpainting and the tasks addressed by us and @cite_1 @cite_17 , is that the previously introduced methods of image inpainting are all in a setting in which the complete ground truth is available. | {
"cite_N": [
"@cite_8",
"@cite_21",
"@cite_1",
"@cite_5",
"@cite_20",
"@cite_13",
"@cite_12",
"@cite_17"
],
"mid": [
"2479644247",
"2963540914",
"2526782364",
"2808402081",
"2784790939",
"345598540",
"2963270367",
"2765732129"
],
"abstract": [
"In this paper, we propose a novel method for image inpainting based on a Deep Convolutional Generative Adversarial Network (DCGAN). We define a loss function consisting of two parts: (1) a contextual loss that preserves similarity between the input corrupted image and the recovered image, and (2) a perceptual loss that ensures a perceptually realistic output image. Given a corrupted image with missing values, we use back-propagation on this loss to map the corrupted image to a smaller latent space. The mapped vector is then passed through the generative model to predict the missing content. The proposed framework is evaluated on the CelebA and SVHN datasets for two challenging inpainting tasks with random 80 corruption and large blocky corruption. Experiments show that our method can successfully predict semantic information in the missing region and achieve pixel-level photorealism, which is impossible by almost all existing methods.",
"Recent deep learning based approaches have shown promising results for the challenging task of inpainting large missing regions in an image. These methods can generate visually plausible image structures and textures, but often create distorted structures or blurry textures inconsistent with surrounding areas. This is mainly due to ineffectiveness of convolutional neural networks in explicitly borrowing or copying information from distant spatial locations. On the other hand, traditional texture and patch synthesis approaches are particularly suitable when it needs to borrow textures from the surrounding regions. Motivated by these observations, we propose a new deep generative model-based approach which can not only synthesize novel image structures but also explicitly utilize surrounding image features as references during network training to make better predictions. The model is a feedforward, fully convolutional neural network which can process images with multiple holes at arbitrary locations and with variable sizes during the test time. Experiments on multiple datasets including faces (CelebA, CelebA-HQ), textures (DTD) and natural images (ImageNet, Places2) demonstrate that our proposed approach generates higher-quality inpainting results than existing ones. Code, demo and models are available at: https: github.com JiahuiYu generative_inpainting.",
"Recently, neuron activations extracted from a pre-trained convolutional neural network (CNN) show promising performance in various visual tasks. However, due to the domain and task bias, using the features generated from the model pre-trained for image classification as image representations for instance retrieval is problematic. In this paper, we propose quartet-net learning to improve the discriminative power of CNN features for instance retrieval. The general idea is to map the features into a space where the image similarity can be better evaluated. Our network differs from the traditional Siamese-net in two ways. First, we adopt a double-margin contrastive loss with a dynamic margin tuning strategy to train the network which leads to more robust performance. Second, we introduce in the mimic learning regularization to improve the generalization ability of the network by preventing it from overfitting to the training data. Catering for the network learning, we collect a large-scale dataset, namely GeoPair, which consists of 68k matching image pairs and 63k non-matching pairs. Experiments on several standard instance retrieval datasets demonstrate the effectiveness of our method.",
"Abstract Although image inpainting is now an effective image editing technique, limited work has been done for inpainting forensics. The main drawbacks of the conventional inpainting forensics methods lie in the difficulties on inpainting feature extraction and the very high computational cost. In this paper, we propose a novel approach based on a convolutional neural network (CNN) to detect patch-based inpainting operation. Specifically, the CNN is built following the encoder–decoder network structure, which allows us to predict the inpainting probability for each pixel in an image. To guide the CNN to automatically learn the inpainting features, a label matrix is generated for the CNN training by assigning a class label for each pixel of an image, and the designed weighted cross-entropy serves as the loss function. They further help to strongly supervise the CNN to capture the manipulation information rather than the image content features. By the established CNN, inpainting forensics does not need to consider feature extraction and classifier design, and use any postprocessing as in conventional forensics methods. They are combined into the unique framework and optimized simultaneously. Experimental results show that the proposed method achieves superior performance in terms of true positive rate, false positive rate and the running time, as compared with state-of-the-art methods for inpainting forensics, and is very robust against JPEG compression and scaling manipulations.",
"Recent deep learning based approaches have shown promising results on image inpainting for the challenging task of filling in large missing regions in an image. These methods can generate visually plausible image structures and textures, but often create distorted structures or blurry textures inconsistent with surrounding areas. This is mainly due to ineffectiveness of convolutional neural networks in explicitly borrowing or copying information from distant spatial locations. On the other hand, traditional texture and patch synthesis approaches are particularly suitable when it needs to borrow textures from the surrounding regions. Motivated by these observations, we propose a new deep generative model-based approach which can not only synthesize novel image structures but also explicitly utilize surrounding image features as references during network training to make better predictions. The model is a feed-forward, fully convolutional neural network which can process images with multiple holes at arbitrary locations and with variable sizes during the test time. Experiments on multiple datasets including faces, textures and natural images demonstrate that the proposed approach generates higher-quality inpainting results than existing ones. Code and trained models will be released.",
"Most inpainting approaches require a good image model to infer the unknown pixels. In this work, we directly learn a mapping from image patches, corrupted by missing pixels, onto complete image patches. This mapping is represented as a deep neural network that is automatically trained on a large image data set. In particular, we are interested in the question whether it is helpful to exploit the shape information of the missing regions, i.e. the masks, which is something commonly ignored by other approaches. In comprehensive experiments on various images, we demonstrate that our learning-based approach is able to use this extra information and can achieve state-of-the-art inpainting results. Furthermore, we show that training with such extra information is useful for blind inpainting, where the exact shape of the missing region might be uncertain, for instance due to aliasing effects.",
"Deep convolutional networks (CNNs) have exhibited their potential in image inpainting for producing plausible results. However, in most existing methods, e.g., context encoder, the missing parts are predicted by propagating the surrounding convolutional features through a fully connected layer, which intends to produce semantically plausible but blurry result. In this paper, we introduce a special shift-connection layer to the U-Net architecture, namely Shift-Net, for filling in missing regions of any shape with sharp structures and fine-detailed textures. To this end, the encoder feature of the known region is shifted to serve as an estimation of the missing parts. A guidance loss is introduced on decoder feature to minimize the distance between the decoder feature after fully connected layer and the ground-truth encoder feature of the missing parts. With such constraint, the decoder feature in missing region can be used to guide the shift of encoder feature in known region. An end-to-end learning algorithm is further developed to train the Shift-Net. Experiments on the Paris StreetView and Places datasets demonstrate the efficiency and effectiveness of our Shift-Net in producing sharper, fine-detailed, and visually plausible results. The codes and pre-trained models are available at https: github.com Zhaoyi-Yan Shift-Net.",
"In recent years, we have witnessed the great success of deep learning on various problems both in low and high-level computer visions. The low-level vision problems, including inpainting, deblurring, denoising, super-resolution, and so on, are highly anticipated to occur in machine vision and image processing. Many deep learning based methods have been proposed to solve low-level vision problems. Most researches treat these problems independently; however, most of the time they appear concurrently. Motivated by the success of generative model in the field of image generation, we develop a deep cascade of neural networks to solve the inpainting, deblurring, denoising problems at the same time. Our model contains two networks: inpainting GAN and deblurring-denoising network. Inpainting GAN generates the coarse patches to fill the lost part in damaged image, and the deblurring-denoising network, stacked by a convolutional auto-encoder, will further refine them. Unlike other methods that handle each problem separately, our method jointly optimizes the two sub-networks. Because GAN training is not only unstable but also difficult, we adopt the Wasserstein distance as the loss function of the inpainting GAN and propose a gradual training strategy. Learning from the idea of residual learning, we utilize skip connections to pass image details from input to reconstruction layer. Experimental results have demonstrated that the proposed model can achieve state-of-the-art performance. Through the experiments, we also demonstrated the effectiveness of the cascade architecture."
]
} |
1907.09786 | 2963372556 | We propose a novel single-step training strategy that allows convolutional encoder-decoder networks that use skip connections, to complete partially observed data by means of hallucination. This strategy is demonstrated for the task of completing 2-D road layouts as well as 3-D vehicle shapes. As input, it takes data from a partially observed domain, for which no ground truth is available, and data from an unpaired prior knowledge domain and trains the network in an end-to-end manner. Our single-step training strategy is compared against two state-of-the-art baselines, one using a two-step auto-encoder training strategy and one using an adversarial strategy. Our novel strategy achieves an improvement up to +12.2 F-measure on the Cityscapes dataset. The learned network intrinsically generalizes better than the baselines on unseen datasets, which is demonstrated by an improvement up to +23.8 F-measure on the unseen KITTI dataset. Moreover, our approach outperforms the baselines using the same backbone network on the 3-D shape completion benchmark by a margin of 0.006 Hamming distance. | If the ground truth is not available, one has to hallucinate the region to be completed. Srikantha and Gall @cite_23 propose a system to hallucinate a depth map and a semantic map, given an RGB image and a noisy, incomplete depth map, which is able to remove the foreground objects. Schulter al @cite_1 proposes a CNN to conduct a similar task without depth, by intentionally adding random foreground masks during training. Recently, a VAE with two-step training @cite_4 is proposed for shape completion. In the first step, a canonical VAE is trained on a complete shape prior dataset which has no direct correspondence to the incomplete shape dataset. Then the amortized maximum likelihood (AML) is applied as supervision on the incomplete shape data with the decoder fixed in the second training step. This VAE approach is originally used to learn 3-D vehicle shape completion, but it can be generalized to similar tasks such as our 2-D road layout hallucinating. Therefore, we use this approach as a baseline, which is referred to as . | {
"cite_N": [
"@cite_4",
"@cite_1",
"@cite_23"
],
"mid": [
"2949257576",
"2585635281",
"2762085884"
],
"abstract": [
"The main contribution of this paper is a simple semi-supervised pipeline that only uses the original training set without collecting extra data. It is challenging in 1) how to obtain more training data only from the training set and 2) how to use the newly generated data. In this work, the generative adversarial network (GAN) is used to generate unlabeled samples. We propose the label smoothing regularization for outliers (LSRO). This method assigns a uniform label distribution to the unlabeled images, which regularizes the supervised model and improves the baseline. We verify the proposed method on a practical problem: person re-identification (re-ID). This task aims to retrieve a query person from other cameras. We adopt the deep convolutional generative adversarial network (DCGAN) for sample generation, and a baseline convolutional neural network (CNN) for representation learning. Experiments show that adding the GAN-generated data effectively improves the discriminative ability of learned CNN embeddings. On three large-scale datasets, Market-1501, CUHK03 and DukeMTMC-reID, we obtain +4.37 , +1.6 and +2.46 improvement in rank-1 precision over the baseline CNN, respectively. We additionally apply the proposed method to fine-grained bird recognition and achieve a +0.6 improvement over a strong baseline. The code is available at this https URL",
"The main contribution of this paper is a simple semisupervised pipeline that only uses the original training set without collecting extra data. It is challenging in 1) how to obtain more training data only from the training set and 2) how to use the newly generated data. In this work, the generative adversarial network (GAN) is used to generate unlabeled samples. We propose the label smoothing regularization for outliers (LSRO). This method assigns a uniform label distribution to the unlabeled images, which regularizes the supervised model and improves the baseline. We verify the proposed method on a practical problem: person re-identification (re-ID). This task aims to retrieve a query person from other cameras. We adopt the deep convolutional generative adversarial network (DCGAN) for sample generation, and a baseline convolutional neural network (CNN) for representation learning. Experiments show that adding the GAN-generated data effectively improves the discriminative ability of learned CNN embeddings. On three large-scale datasets, Market- 1501, CUHK03 and DukeMTMC-reID, we obtain +4.37 , +1.6 and +2.46 improvement in rank-1 precision over the baseline CNN, respectively. We additionally apply the proposed method to fine-grained bird recognition and achieve a +0.6 improvement over a strong baseline. The code is available at https: github.com layumi Person-reID_GAN.",
"Sufficient training examples are the fundamental requirement for most of the learning tasks. However, collecting well-labelled training examples is costly. Inspired by Zero-shot Learning (ZSL) that can make use of visual attributes or natural language semantics as an intermediate level clue to associate low-level features with high-level classes, in a novel extension of this idea, we aim to synthesise training data for novel classes using only semantic attributes. Despite the simplicity of this idea, there are several challenges. First, how to prevent the synthesised data from over-fitting to training classes? Second, how to guarantee the synthesised data is discriminative for ZSL tasks? Third, we observe that only a few dimensions of the learnt features gain high variances whereas most of the remaining dimensions are not informative. Thus, the question is how to make the concentrated information diffuse to most of the dimensions of synthesised data. To address the above issues, we propose a novel embedding algorithm named Unseen Visual Data Synthesis (UVDS) that projects semantic features to the high-dimensional visual feature space. Two main techniques are introduced in our proposed algorithm. (1) We introduce a latent embedding space which aims to reconcile the structural difference between the visual and semantic spaces, meanwhile preserve the local structure. (2) We propose a novel Diffusion Regularisation (DR) that explicitly forces the variances to diffuse over most dimensions of the synthesised data. By an orthogonal rotation (more precisely, an orthogonal transformation), DR can remove the redundant correlated attributes and further alleviate the over-fitting problem. On four benchmark datasets, we demonstrate the benefit of using synthesised unseen data for zero-shot learning. Extensive experimental results suggest that our proposed approach significantly outperforms the state-of-the-art methods."
]
} |
1907.09786 | 2963372556 | We propose a novel single-step training strategy that allows convolutional encoder-decoder networks that use skip connections, to complete partially observed data by means of hallucination. This strategy is demonstrated for the task of completing 2-D road layouts as well as 3-D vehicle shapes. As input, it takes data from a partially observed domain, for which no ground truth is available, and data from an unpaired prior knowledge domain and trains the network in an end-to-end manner. Our single-step training strategy is compared against two state-of-the-art baselines, one using a two-step auto-encoder training strategy and one using an adversarial strategy. Our novel strategy achieves an improvement up to +12.2 F-measure on the Cityscapes dataset. The learned network intrinsically generalizes better than the baselines on unseen datasets, which is demonstrated by an improvement up to +23.8 F-measure on the unseen KITTI dataset. Moreover, our approach outperforms the baselines using the same backbone network on the 3-D shape completion benchmark by a margin of 0.006 Hamming distance. | Road layout hallucinating can be seen as a specific approach of road layout understanding, which is an important task for robot and intelligent vehicle navigation. One challenge is that the ego-centric sensory data usually contains occlusions of the foreground objects, which makes roads visually incomplete. Many works tackling occlusions are focusing on using front-view images, such as road boundary detection @cite_9 , and road segmentation @cite_22 . Less work has been carried out on the top-view ego-centric sensing. In @cite_1 , the proposed system can produce a top-view road layout, while the occlusion is still addressed on the front-view image by pre-processing. In the later top-view refinement, the GPS is heavily relied on for a paired reconstruction supervision. We use a variant of this method, which only uses unpaired prior knowledge and thus no GPS pairing, as the second baseline. It is referred to as . | {
"cite_N": [
"@cite_9",
"@cite_1",
"@cite_22"
],
"mid": [
"2964294967",
"2769967426",
"2167222293"
],
"abstract": [
"Given a single RGB image of a complex outdoor road scene in the perspective view, we address the novel problem of estimating an occlusion-reasoned semantic scene layout in the top-view. This challenging problem not only requires an accurate understanding of both the 3D geometry and the semantics of the visible scene, but also of occluded areas. We propose a convolutional neural network that learns to predict occluded portions of the scene layout by looking around foreground objects like cars or pedestrians. But instead of hallucinating RGB values, we show that directly predicting the semantics and depths in the occluded areas enables a better transformation into the top-view. We further show that this initial top-view representation can be significantly enhanced by learning priors and rules about typical road layouts from simulated or, if available, map data. Crucially, training our model does not require costly or subjective human annotations for occluded areas or the top-view, but rather uses readily available annotations for standard semantic segmentation in the perspective view. We extensively evaluate and analyze our approach on the KITTI and Cityscapes data sets.",
"We present a self-supervised approach to ignoring \"distractors\" in camera images for the purposes of robustly estimating vehicle motion in cluttered urban environments. We leverage offline multi-session mapping approaches to automatically generate a per-pixel ephemerality mask and depth map for each input image, which we use to train a deep convolutional network. At run-time we use the predicted ephemerality and depth as an input to a monocular visual odometry (VO) pipeline, using either sparse features or dense photometric matching. Our approach yields metric-scale VO using only a single camera and can recover the correct egomotion even when 90 of the image is obscured by dynamic, independently moving objects. We evaluate our robust VO methods on more than 400km of driving from the Oxford RobotCar Dataset and demonstrate reduced odometry drift and significantly improved egomotion estimation in the presence of large moving vehicles in urban traffic.",
"Detecting the road area and ego-lane ahead of a vehicle is central to modern driver assistance systems. While lane-detection on well-marked roads is already available in modern vehicles, finding the boundaries of unmarked or weakly marked roads and lanes as they appear in inner-city and rural environments remains an unsolved problem due to the high variability in scene layout and illumination conditions, amongst others. While recent years have witnessed great interest in this subject, to date no commonly agreed upon benchmark exists, rendering a fair comparison amongst methods difficult. In this paper, we introduce a novel open-access dataset and benchmark for road area and ego-lane detection. Our dataset comprises 600 annotated training and test images of high variability from the KITTI autonomous driving project, capturing a broad spectrum of urban road scenes. For evaluation, we propose to use the 2D Bird's Eye View (BEV) space as vehicle control usually happens in this 2D world, requiring detection results to be represented in this very same space. Furthermore, we propose a novel, behavior-based metric which judges the utility of the extracted ego-lane area for driver assistance applications by fitting a driving corridor to the road detection results in the BEV. We believe this to be important for a meaningful evaluation as pixel-level performance is of limited value for vehicle control. State-of-the-art road detection algorithms are used to demonstrate results using classical pixel-level metrics in perspective and BEV space as well as the novel behavior-based performance measure. All data and annotations are made publicly available on the KITTI online evaluation website in order to serve as a common benchmark for road terrain detection algorithms."
]
} |
1907.09815 | 2963714798 | The interaction between language and visual information has been emphasized in visual question answering (VQA) with the help of attention mechanism. However, the relationship between words in question has been underestimated, which makes it hard to answer questions that involve the relationship between multiple entities, such as comparison and counting. In this paper, we develop the graph reasoning networks to tackle this problem. Two kinds of graphs are investigated, namely inter-graph and intra-graph. The inter-graph transfers features of the detected objects to their related query words, enabling the output nodes to have both semantic and factual information. The intra-graph exchanges information between these output nodes from inter-graph to amplify implicit yet important relationship between objects. These two kinds of graphs cooperate with each other, and thus our resulting model can reason the relationship and dependence between objects, which leads to realization of multi-step reasoning. Experimental results on the GQA v1.1 dataset demonstrate the reasoning ability of our method to handle compositional questions about real-world images. We achieve state-of-the-art performance, boosting accuracy to 57.04 . On the VQA 2.0 dataset, we also receive a promising improvement on overall accuracy, especially on counting problem. | : VQA is a task to answer the given question based on the input image. The question is usually embedded into a vector with LSTM @cite_6 , and the image is represented by the fixed-size grid features from ResNet @cite_4 . Recently, @cite_28 focuses on bottom-up attention of image features and proposes a set of salient image regions with natural expression and additional attributes detected by Faster-RCNN @cite_3 . Furthermore, its training set contains 1,600 object classes and 400 attribute classes, larger than the original 80 object classes. | {
"cite_N": [
"@cite_28",
"@cite_3",
"@cite_4",
"@cite_6"
],
"mid": [
"2949980205",
"2522258376",
"2770883544",
"2950761309"
],
"abstract": [
"Visual Question Answering (VQA) is the task of taking as input an image and a free-form natural language question about the image, and producing an accurate answer. In this work we view VQA as a \"feature extraction\" module to extract image and caption representations. We employ these representations for the task of image-caption ranking. Each feature dimension captures (imagines) whether a fact (question-answer pair) could plausibly be true for the image and caption. This allows the model to interpret images and captions from a wide variety of perspectives. We propose score-level and representation-level fusion models to incorporate VQA knowledge in an existing state-of-the-art VQA-agnostic image-caption ranking model. We find that incorporating and reasoning about consistency between images and captions significantly improves performance. Concretely, our model improves state-of-the-art on caption retrieval by 7.1 and on image retrieval by 4.4 on the MSCOCO dataset.",
"This paper proposes to improve visual question answering (VQA) with structured representations of both scene contents and questions. A key challenge in VQA is to require joint reasoning over the visual and text domains. The predominant CNN LSTM-based approach to VQA is limited by monolithic vector representations that largely ignore structure in the scene and in the form of the question. CNN feature vectors cannot effectively capture situations as simple as multiple object instances, and LSTMs process questions as series of words, which does not reflect the true complexity of language structure. We instead propose to build graphs over the scene objects and over the question words, and we describe a deep neural network that exploits the structure in these representations. This shows significant benefit over the sequential processing of LSTMs. The overall efficacy of our approach is demonstrated by significant improvements over the state-of-the-art, from 71.2 to 74.4 in accuracy on the \"abstract scenes\" multiple-choice benchmark, and from 34.7 to 39.1 in accuracy over pairs of \"balanced\" scenes, i.e. images with fine-grained differences and opposite yes no answers to a same question.",
"Recently, the Visual Question Answering (VQA) task has gained increasing attention in artificial intelligence. Existing VQA methods mainly adopt the visual attention mechanism to associate the input question with corresponding image regions for effective question answering. The free-form region based and the detection-based visual attention mechanisms are mostly investigated, with the former ones attending free-form image regions and the latter ones attending pre-specified detection-box regions. We argue that the two attention mechanisms are able to provide complementary information and should be effectively integrated to better solve the VQA problem. In this paper, we propose a novel deep neural network for VQA that integrates both attention mechanisms. Our proposed framework effectively fuses features from free-form image regions, detection boxes, and question representations via a multi-modal multiplicative feature embedding scheme to jointly attend question-related free-form image regions and detection boxes for more accurate question answering. The proposed method is extensively evaluated on two publicly available datasets, COCO-QA and VQA, and outperforms state-of-the-art approaches. Source code is available at this https URL",
"We propose the task of free-form and open-ended Visual Question Answering (VQA). Given an image and a natural language question about the image, the task is to provide an accurate natural language answer. Mirroring real-world scenarios, such as helping the visually impaired, both the questions and answers are open-ended. Visual questions selectively target different areas of an image, including background details and underlying context. As a result, a system that succeeds at VQA typically needs a more detailed understanding of the image and complex reasoning than a system producing generic image captions. Moreover, VQA is amenable to automatic evaluation, since many open-ended answers contain only a few words or a closed set of answers that can be provided in a multiple-choice format. We provide a dataset containing 0.25M images, 0.76M questions, and 10M answers (www.visualqa.org), and discuss the information it provides. Numerous baselines and methods for VQA are provided and compared with human performance. Our VQA demo is available on CloudCV (this http URL)."
]
} |
1907.09815 | 2963714798 | The interaction between language and visual information has been emphasized in visual question answering (VQA) with the help of attention mechanism. However, the relationship between words in question has been underestimated, which makes it hard to answer questions that involve the relationship between multiple entities, such as comparison and counting. In this paper, we develop the graph reasoning networks to tackle this problem. Two kinds of graphs are investigated, namely inter-graph and intra-graph. The inter-graph transfers features of the detected objects to their related query words, enabling the output nodes to have both semantic and factual information. The intra-graph exchanges information between these output nodes from inter-graph to amplify implicit yet important relationship between objects. These two kinds of graphs cooperate with each other, and thus our resulting model can reason the relationship and dependence between objects, which leads to realization of multi-step reasoning. Experimental results on the GQA v1.1 dataset demonstrate the reasoning ability of our method to handle compositional questions about real-world images. We achieve state-of-the-art performance, boosting accuracy to 57.04 . On the VQA 2.0 dataset, we also receive a promising improvement on overall accuracy, especially on counting problem. | Based on the fusion methods of the two features, we can classify VQA models into two categories: early fusion models and later fusion models. Early fusion models try to fine-tune the image classification network with the intervention of the question, they insert the question embedding into the batch normalization layer @cite_23 @cite_31 to propose MODERN architecture. These models have less risk of over-fitting because of affecting less than 1 However, sometimes, the information given in the image is not enough to infer the right answer, common sense is required in the external knowledge-based models. @cite_9 builds the FVQA dataset based on DBpedia @cite_30 , ConceptNet @cite_14 and WebChild @cite_34 . @cite_1 queries the triplet (visual concept, relation, attribute) in this dataset to score the retrieved facts. And @cite_32 builds a relation graph based on the retrieved facts regarding the visual concept and attribute as nodes and relation as links to exchange information. | {
"cite_N": [
"@cite_30",
"@cite_14",
"@cite_9",
"@cite_1",
"@cite_32",
"@cite_23",
"@cite_31",
"@cite_34"
],
"mid": [
"2962980263",
"2771951981",
"2799345082",
"2174492417",
"2963150162",
"2747148117",
"2952228917",
"2744822616"
],
"abstract": [
"Visual question answering (VQA) is challenging, because it requires a simultaneous understanding of both visual content of images and textual content of questions. To support the VQA task, we need to find good solutions for the following three issues: 1) fine-grained feature representations for both the image and the question; 2) multimodal feature fusion that is able to capture the complex interactions between multimodal features; and 3) automatic answer prediction that is able to consider the complex correlations between multiple diverse answers for the same question. For fine-grained image and question representations, a “coattention” mechanism is developed using a deep neural network (DNN) architecture to jointly learn the attentions for both the image and the question, which can allow us to reduce the irrelevant features effectively and obtain more discriminative features for image and question representations. For multimodal feature fusion, a generalized multimodal factorized high-order pooling approach (MFH) is developed to achieve more effective fusion of multimodal features by exploiting their correlations sufficiently, which can further result in superior VQA performance as compared with the state-of-the-art approaches. For answer prediction, the Kullback–Leibler divergence is used as the loss function to achieve precise characterization of the complex correlations between multiple diverse answers with the same or similar meaning, which can allow us to achieve faster convergence rate and obtain slightly better accuracy on answer prediction. A DNN architecture is designed to integrate all these aforementioned modules into a unified model for achieving superior VQA performance. With an ensemble of our MFH models, we achieve the state-of-the-art performance on the large-scale VQA data sets and win the runner-up in VQA Challenge 2017.",
"A number of studies have found that today's Visual Question Answering (VQA) models are heavily driven by superficial correlations in the training data and lack sufficient image grounding. To encourage development of models geared towards the latter, we propose a new setting for VQA where for every question type, train and test sets have different prior distributions of answers. Specifically, we present new splits of the VQA v1 and VQA v2 datasets, which we call Visual Question Answering under Changing Priors (VQA-CP v1 and VQA-CP v2 respectively). First, we evaluate several existing VQA models under this new setting and show that their performance degrades significantly compared to the original VQA setting. Second, we propose a novel Grounded Visual Question Answering model (GVQA) that contains inductive biases and restrictions in the architecture specifically designed to prevent the model from 'cheating' by primarily relying on priors in the training data. Specifically, GVQA explicitly disentangles the recognition of visual concepts present in the image from the identification of plausible answer space for a given question, enabling the model to more robustly generalize across different distributions of answers. GVQA is built off an existing VQA model -- Stacked Attention Networks (SAN). Our experiments demonstrate that GVQA significantly outperforms SAN on both VQA-CP v1 and VQA-CP v2 datasets. Interestingly, it also outperforms more powerful VQA models such as Multimodal Compact Bilinear Pooling (MCB) in several cases. GVQA offers strengths complementary to SAN when trained and evaluated on the original VQA v1 and VQA v2 datasets. Finally, GVQA is more transparent and interpretable than existing VQA models.",
"Existing attention mechanisms either attend to local image grid or object level features for Visual Question Answering (VQA). Motivated by the observation that questions can relate to both object instances and their parts, we propose a novel attention mechanism that jointly considers reciprocal relationships between the two levels of visual details. The bottom-up attention thus generated is further coalesced with the top-down information to only focus on the scene elements that are most relevant to a given question. Our design hierarchically fuses multi-modal information i.e., language, object- and gird-level features, through an efficient tensor decomposition scheme. The proposed model improves the state-of-the-art single model performances from 67.9 to 68.2 on VQAv1 and from 65.3 to 67.4 on VQAv2, demonstrating a significant boost.",
"We propose a novel attention based deep learning architecture for visual question answering task (VQA). Given an image and an image related natural language question, VQA generates the natural language answer for the question. Generating the correct answers requires the model's attention to focus on the regions corresponding to the question, because different questions inquire about the attributes of different image regions. We introduce an attention based configurable convolutional neural network (ABC-CNN) to learn such question-guided attention. ABC-CNN determines an attention map for an image-question pair by convolving the image feature map with configurable convolutional kernels derived from the question's semantics. We evaluate the ABC-CNN architecture on three benchmark VQA datasets: Toronto COCO-QA, DAQUAR, and VQA dataset. ABC-CNN model achieves significant improvements over state-of-the-art methods on these datasets. The question-guided attention generated by ABC-CNN is also shown to reflect the regions that are highly relevant to the questions.",
"Visual question answering (VQA) is challenging because it requires a simultaneous understanding of both the visual content of images and the textual content of questions. The approaches used to represent the images and questions in a fine-grained manner and questions and to fuse these multimodal features play key roles in performance. Bilinear pooling based models have been shown to outperform traditional linear models for VQA, but their high-dimensional representations and high computational complexity may seriously limit their applicability in practice. For multimodal feature fusion, here we develop a Multi-modal Factorized Bilinear (MFB) pooling approach to efficiently and effectively combine multi-modal features, which results in superior performance for VQA compared with other bilinear pooling approaches. For fine-grained image and question representation, we develop a ‘co-attention’ mechanism using an end-to-end deep network architecture to jointly learn both the image and question attentions. Combining the proposed MFB approach with co-attention learning in a new network architecture provides a unified model for VQA. Our experimental results demonstrate that the single MFB with co-attention model achieves new state-of-theart performance on the real-world VQA dataset. Code available at https: github.com yuzcccc mfb.",
"Rich and dense human labeled datasets are among the main enabling factors for the recent advance on vision-language understanding. Many seemingly distant annotations (e.g., semantic segmentation and visual question answering (VQA)) are inherently connected in that they reveal different levels and perspectives of human understandings about the same visual scenes --- and even the same set of images (e.g., of COCO). The popularity of COCO correlates those annotations and tasks. Explicitly linking them up may significantly benefit both individual tasks and the unified vision and language modeling. We present the preliminary work of linking the instance segmentations provided by COCO to the questions and answers (QAs) in the VQA dataset, and name the collected links visual questions and segmentation answers (VQS). They transfer human supervision between the previously separate tasks, offer more effective leverage to existing problems, and also open the door for new research problems and models. We study two applications of the VQS data in this paper: supervised attention for VQA and a novel question-focused semantic segmentation task. For the former, we obtain state-of-the-art results on the VQA real multiple-choice task by simply augmenting the multilayer perceptrons with some attention features that are learned using the segmentation-QA links as explicit supervision. To put the latter in perspective, we study two plausible methods and compare them to an oracle method assuming that the instance segmentations are given at the test stage.",
"Problems at the intersection of vision and language are of significant importance both as challenging research questions and for the rich set of applications they enable. However, inherent structure in our world and bias in our language tend to be a simpler signal for learning than visual modalities, resulting in models that ignore visual information, leading to an inflated sense of their capability. We propose to counter these language priors for the task of Visual Question Answering (VQA) and make vision (the V in VQA) matter! Specifically, we balance the popular VQA dataset by collecting complementary images such that every question in our balanced dataset is associated with not just a single image, but rather a pair of similar images that result in two different answers to the question. Our dataset is by construction more balanced than the original VQA dataset and has approximately twice the number of image-question pairs. Our complete balanced dataset is publicly available at this http URL as part of the 2nd iteration of the Visual Question Answering Dataset and Challenge (VQA v2.0). We further benchmark a number of state-of-art VQA models on our balanced dataset. All models perform significantly worse on our balanced dataset, suggesting that these models have indeed learned to exploit language priors. This finding provides the first concrete empirical evidence for what seems to be a qualitative sense among practitioners. Finally, our data collection protocol for identifying complementary images enables us to develop a novel interpretable model, which in addition to providing an answer to the given (image, question) pair, also provides a counter-example based explanation. Specifically, it identifies an image that is similar to the original image, but it believes has a different answer to the same question. This can help in building trust for machines among their users.",
"Visual question answering (VQA) is challenging because it requires a simultaneous understanding of both the visual content of images and the textual content of questions. The approaches used to represent the images and questions in a fine-grained manner and questions and to fuse these multi-modal features play key roles in performance. Bilinear pooling based models have been shown to outperform traditional linear models for VQA, but their high-dimensional representations and high computational complexity may seriously limit their applicability in practice. For multi-modal feature fusion, here we develop a Multi-modal Factorized Bilinear (MFB) pooling approach to efficiently and effectively combine multi-modal features, which results in superior performance for VQA compared with other bilinear pooling approaches. For fine-grained image and question representation, we develop a co-attention mechanism using an end-to-end deep network architecture to jointly learn both the image and question attentions. Combining the proposed MFB approach with co-attention learning in a new network architecture provides a unified model for VQA. Our experimental results demonstrate that the single MFB with co-attention model achieves new state-of-the-art performance on the real-world VQA dataset. Code available at this https URL."
]
} |
1907.09871 | 2963864879 | The paper details the first successful attempt at using model-checking techniques to verify the correctness of distributed algorithms for robots evolving in a environment. The study focuses on the problem of rendezvous of two robots with lights. There exist many different rendezvous algorithms that aim at finding the minimal number of colors needed to solve rendezvous in various synchrony models (e.g., FSYNC, SSYNC, ASYNC). While these rendezvous algorithms are typically very simple, their analysis and proof of correctness tend to be extremely complex, tedious, and error-prone as impossibility results are based on subtle interactions between robots activation schedules. The paper presents a generic verification model written for the SPIN model-checker. In particular, we explain the subtle design decisions that allow to keep the search space finite and tractable, as well as prove several important theorems that support them. As a sanity check, we use the model to verify several known rendezvous algorithms in six different models of synchrony. In each case, we find that the results obtained from the model-checker are consistent with the results known in the literature. The model-checker outputs a counter-example execution in every case that is known to fail. In the course of developing and proving the validity of the model, we identified several fundamental theorems, including the ability for a well chosen algorithm and ASYNC scheduler to produce an emerging property of memory in a system of oblivious mobile robots, and why it is not a problem for luminous rendezvous algorithms. | Designing and proving mobile robot protocols is notoriously difficult. Formal methods encompass a long-lasting path of research that is meant to overcome errors of human origin. Unsurprisingly, this mechanized approach to protocol correctness was used in the context of mobile robots @cite_11 @cite_32 @cite_5 @cite_18 @cite_1 @cite_27 @cite_15 @cite_28 @cite_32 @cite_21 @cite_2 . | {
"cite_N": [
"@cite_18",
"@cite_28",
"@cite_21",
"@cite_1",
"@cite_32",
"@cite_27",
"@cite_2",
"@cite_5",
"@cite_15",
"@cite_11"
],
"mid": [
"2338820189",
"2782960731",
"2054245952",
"1477234322",
"1574773379",
"2011767579",
"1995782858",
"2752666516",
"2963688871",
"2130917470"
],
"abstract": [
"Mobile robot networks emerged in the past few years as a promising distributed computing model. Existing work in the literature typically ensures the correctness of mobile robot protocols via ad hoc handwritten proofs, which, in the case of asynchronous execution models, are both cumbersome and error-prone. Our contribution is twofold. We first propose a formal model to describe mobile robot protocols operating in a discrete space i.e., with a finite set of possible robot positions, under synchrony and asynchrony assumptions. We translate this formal model into the DVE language, which is the input format of the model-checkers DiVinE and ITS tools, and formally prove the equivalence of the two models. We then verify several instances of two existing protocols for variants of the ring exploration in an asynchronous setting: exploration with stop and perpetual exclusive exploration. For the first protocol we refine the correctness bounds and for the second one, we exhibit a counter-example. This protocol is then modified and we establish the correctness of the new version with an inductive proof.",
"Swarms of mobile robots recently attracted the focus of the Distributed Computing community. One of the fundamental problems in this context is that of exploration: the robots must coordinate to visit all locations that are reachable from their initial positions. Despite its apparent simplicity, this problem proved quite hard to characterise fully, due to many model variants, leading to informal error-prone reasoning. Over the past few years, a significant effort permitted to set up a formal framework, relying on the Coq proof assistant, which was used to provide certified results when robots evolve in a continuous bi-dimensional Euclidean space. However, the most challenging issues with exploration arise in the discrete setting (a.k.a. graph), where locations are modeled as vertices and where edges between vertices denote the ability for a robot to move from one location to the next. We present a formal model to tackle problems and reason about robot algorithms arising in the discrete setting. Our approach extends and generalises previous research efforts focusing on the continuous model. As case studies, we consider fundamental impossibility results for exploration with stop in the discrete model. To our knowledge, those are the first certified results in this context. This framework paves the way for a general certification workflow dedicated to mobile robots on graphs.",
"Navigating and path planning in environments with limited a priori knowledge is a fundamental challenge for mobile robots. Robots operating in human-occupied environments must also respect sociocontextual boundaries such as personal workspaces. There is a need for robots to be able to navigate in such environments without having to explore and build an intricate representation of the world. In this paper, a method for supplementing directly observed environmental information with indirect observations of occupied space is presented. The proposed approach enables the online inclusion of novel human positional traces and environment information into a probabilistic framework for path planning. Encapsulation of sociocontextual information, such as identifying areas that people tend to use to move through the environment, is inherently achieved without supervised learning or labelling. Our method bootstraps navigation with indirectly observed sensor data, and leverages the flexibility of the Gaussian process (GP) for producing a navigational map that sampling based path planers such as Probabilistic Roadmaps (PRM) can effectively utilise. Empirical results on a mobile platform demonstrate that a robot can efficiently and socially-appropriately reach a desired goal by exploiting the navigational map in our Bayesian statistical framework.",
"In this paper, we propose a flexible system for robust natural language interpretation of spoken commands on a mobile robot in domestic service robotics applications. Existing language processing for instructing a mobile robot is often restricted by using a simple grammar where precisely pre-defined utterances are directly mapped to system calls. These approaches do not regard fallibility of human users and they only allow for binary processing of an utterance; either a command is part of the grammar and hence understood correctly, or it is not part of the grammar and gets rejected. We model the language processing as an interpretation process where the utterance needs to be mapped to the robot’s capabilities. We do so by casting the processing as a (decision-theoretic) planning problem on interpretation actions. This allows for a flexible system that can resolve ambiguities and which is also capable of initiating steps to achieve clarification. We show how we evaluated several versions of the system with multiple utterances of different complexity as well as with incomplete and erroneous requests.",
"We demonstrate distributed, online, and real-time cooperative localization and mapping between multiple robots operating throughout an unknown environment using indirect measurements. We present a novel Expectation Maximization (EM) based approach to efficiently identify inlier multi-robot loop closures by incorporating robot pose uncertainty, which significantly improves the trajectory accuracy over long-term navigation. An EM and hypothesis based method is used to determine a common reference frame. We detail a 2D laser scan correspondence method to form robust correspondences between laser scans shared amongst robots. The implementation is experimentally validated using teams of aerial vehicles, and analyzed to determine its accuracy, computational efficiency, scalability to many robots, and robustness to varying environments. We demonstrate through multiple experiments that our method can efficiently build maps of large indoor and outdoor environments in a distributed, online, and real-time setting.",
"Long-term autonomous mobile robot operation requires considering place recognition decisions with great caution. A single incorrect decision that is not detected and reconsidered can corrupt the environment model that the robot is trying to build and maintain. This work describes a consensus-based approach to robust place recognition over time, that takes into account all the available information to detect and remove past incorrect loop closures. The main novelties of our work are: (1) the ability of realizing that, in light of new evidence, an incorrect past loop closing decision has been made; the incorrect information can be removed thus recovering the correct estimation with a novel algorithm; (2) extending our proposal to incremental operation; and (3) handling multi-session, spatially related or unrelated scenarios in a unified manner. We demonstrate our proposal, the RRR algorithm, on different odometry systems, e.g. visual or laser, using different front-end loop-closing techniques. For our experiments we use the efficient graph optimization framework g2o as back-end. We back our claims up with several experiments carried out on real data, in single and multi-session experiments showing better results than those obtained by state-of-the-art methods, comparisons against whom are also presented.",
"When humans and mobile robots share the same space, one of their challenges is to navigate around each other and manage their mutual navigational intents. While humans have developed excellent skills in inferring their counterpart's intentions via a number of implicit and non-verbal cues, making navigation also in crowds an ease, this kind of effective and efficient communication often falls short in human-robot encounters. In this paper, two alternative approaches to convey navigational intent of a mobile robot to humans in a shared environment are proposed and analysed. The first is utilising anthropomorphic features of the mobile robot to realise an implicit joint attention using gaze to represent the direction of navigational intent. In the second approach, a more technical design adopting the semantics of car's turn indicators, has been implemented. The paper compares both approaches with each other and against a control behaviour without any communication of intent. Both approaches show statistically significant differences in comparison to the control behaviour. However, the second approach using indicators has shown as being more effective in conveying the intent and also has a higher positive impact on the comfort of the humans encountering the robot.",
"A major challenge towards large scale deployment of autonomous mobile robots is to program them with formal guarantees and high assurance of correct operation. To this end, we present a framework for building safe robots. Our approach for validating the end-to-end correctness of robotics system consists of two parts: (1) a high-level programming language for implementing and systematically testing the reactive robotics software via model checking; (2) a signal temporal logic (STL) based online monitoring system to ensure that the assumptions about the low-level controllers (discrete models) used during model checking hold at runtime. Combining model checking with runtime verification helps us bridge the gap between software verification (discrete) that makes assumptions about the low-level controllers and the physical world, and the actual execution of the software on a real robotic platform in the physical world. To demonstrate the efficacy of our approach, we build a safe adaptive surveillance system and present software-in-the-loop simulations of the application.",
"We study verification problems for autonomous swarms of mobile robots that self-organize and cooperate to solve global objectives. In particular, we focus in this paper on the model proposed by Suzuki and Yamashita of anonymous robots evolving in a discrete space with a finite number of locations (here, a ring). A large number of algorithms have been proposed working for rings whose size is not a priori fixed and can be hence considered as a parameter. Handmade correctness proofs of these algorithms have been shown to be error-prone, and recent attention had been given to the application of formal methods to automatically prove those. Our work is the first to study the verification problem of such algorithms in the parameterized case. We show that safety and reachability problems are undecidable for robots evolving asynchronously. On the positive side, we show that safety properties are decidable in the synchronous case, as well as in the asynchronous case for a particular class of algorithms. Several properties on the protocol can be decided as well. Decision procedures rely on an encoding in Presburger arithmetics formulae that can be verified by an SMT-solver. Feasibility of our approach is demonstrated by the encoding of several case studies.",
"An important competence for a mobile robot system is the ability to localize and perform context interpretation. This is required to perform basic navigation and to facilitate local specific services. Usually localization is performed based on a purely geometric model. Through use of vision and place recognition a number of opportunities open up in terms of flexibility and association of semantics to the model. To achieve this we present an appearance based method for place recognition. The method is based on a large margin classifier in combination with a rich global image descriptor. The method is robust to variations in illumination and minor scene changes. The method is evaluated across several different cameras, changes in time-of-day and weather conditions. The results clearly demonstrate the value of the approach."
]
} |
1907.09871 | 2963864879 | The paper details the first successful attempt at using model-checking techniques to verify the correctness of distributed algorithms for robots evolving in a environment. The study focuses on the problem of rendezvous of two robots with lights. There exist many different rendezvous algorithms that aim at finding the minimal number of colors needed to solve rendezvous in various synchrony models (e.g., FSYNC, SSYNC, ASYNC). While these rendezvous algorithms are typically very simple, their analysis and proof of correctness tend to be extremely complex, tedious, and error-prone as impossibility results are based on subtle interactions between robots activation schedules. The paper presents a generic verification model written for the SPIN model-checker. In particular, we explain the subtle design decisions that allow to keep the search space finite and tractable, as well as prove several important theorems that support them. As a sanity check, we use the model to verify several known rendezvous algorithms in six different models of synchrony. In each case, we find that the results obtained from the model-checker are consistent with the results known in the literature. The model-checker outputs a counter-example execution in every case that is known to fail. In the course of developing and proving the validity of the model, we identified several fundamental theorems, including the ability for a well chosen algorithm and ASYNC scheduler to produce an emerging property of memory in a system of oblivious mobile robots, and why it is not a problem for luminous rendezvous algorithms. | When robots move freely in a continuous two-dimensional Euclidean space (as considered in this paper), to the best of our knowledge the only formal framework available is Pactole. http: pactole.lri.fr It relies on higher-order logic to certify impossibility results @cite_18 @cite_27 @cite_2 , as well as the correctness of algorithms @cite_25 @cite_32 in the and models, possibly for an arbitrary number of robots (hence in a scalable manner). Pactole was recently extended by Balabonski al @cite_29 to handle the model, thanks to its modular design. However, in its current form, Pactole lacks automation; that is, in order to prove a result formally, one still has to write the proof (that is automatically verified), which requires expertise both in Coq (the language Pactole is based upon) and about the mathematical and logical arguments one should use to complete the proof. | {
"cite_N": [
"@cite_18",
"@cite_29",
"@cite_32",
"@cite_27",
"@cite_2",
"@cite_25"
],
"mid": [
"2400422553",
"1750856813",
"2049232787",
"2569270512",
"2963730940",
"2963688871"
],
"abstract": [
"Consider a set of mobile robots placed on distinct nodes of a discrete, anonymous, and bidirectional ring. Asynchronously, each robot takes a snapshot of the ring, determining the size of the ring and which nodes are either occupied by robots or empty. Based on the observed configuration, it decides whether to move to one of its adjacent nodes or not. In the first case, it performs the computed move, eventually. This model of computation is known as Look-Compute-Move. The computation depends on the required task. In this paper, we solve both the well-known Gathering and Exclusive Searching tasks. In the former problem, all robots must simultaneously occupy the same node, eventually. In the latter problem, the aim is to clear all edges of the graph. An edge is cleared if it is traversed by a robot or if both its endpoints are occupied. We consider the exclusive searching where it must be ensured that two robots never occupy the same node. Moreover, since the robots are oblivious, the clearing is perpetual, i.e., the ring is cleared infinitely often. In the literature, most contributions are restricted to a subset of initial configurations. Here, we design two different algorithms and provide a characterization of the initial configurations that permit the resolution of the problems under very weak assumptions. More precisely, we provide a full characterization (except for few pathological cases) of the initial configurations for which gathering can be solved. The algorithm relies on the necessary assumption of the local-weak multiplicity detection. This means that during the Look phase a robot detects also whether the node it occupies is occupied by other robots, without acquiring the exact number. For the exclusive searching, we characterize all (except for few pathological cases) aperiodic configurations from which the problem is feasible. We also provide some impossibility results for the case of periodic configurations.",
"We propose a framework to build formal developments for robot networks using the Coq proof assistant, to state and prove formally various properties. We focus in this paper on impossibility proofs, as it is natural to take advantage of the Coq higher order calculus to reason about algorithms as abstract objects. We present in particular formal proofs of two impossibility results for convergence of oblivious mobile robots if respectively more than one half and more than one third of the robots exhibit Byzantine failures, starting from the original theorems by . Thanks to our formalisation, the corresponding Coq developments are quite compact. To our knowledge, these are the first certified (in the sense of formally proved) impossibility results for robot networks.",
"Anonymous mobile robots are often classified into synchronous, semi-synchronous, and asynchronous robots when discussing the pattern formation problem. For semi-synchronous robots, all patterns formable with memory are also formable without memory, with the single exception of forming a point (i.e., the gathering) by two robots. (All patterns formable with memory are formable without memory for synchronous robots, and little is known for asynchronous robots.) However, the gathering problem for two semi-synchronous robots without memory (called oblivious robots in this paper) is trivially solvable when their local coordinate systems are consistent, and the impossibility proof essentially uses the inconsistencies in their coordinate systems. Motivated by this, this paper investigates the magnitude of consistency between the local coordinate systems necessary and sufficient to solve the gathering problem for two oblivious robots under semi-synchronous and asynchronous models. To discuss the magnitude of consistency, we assume that each robot is equipped with an unreliable compass, the bearings of which may deviate from an absolute reference direction, and that the local coordinate system of each robot is determined by its compass. We consider two families of unreliable compasses, namely, static compasses with (possibly incorrect) constant bearings and dynamic compasses the bearings of which can change arbitrarily (immediately before a new look-compute-move cycle starts and after the last cycle ends). For each of the combinations of robot and compass models, we establish the condition on deviation @math that allows an algorithm to solve the gathering problem, where the deviation is measured by the largest angle formed between the @math -axis of a compass and the reference direction of the global coordinate system: @math for semi-synchronous and asynchronous robots with static compasses, @math for semi-synchronous robots with dynamic compasses, and @math for asynchronous robots with dynamic compasses. Except for asynchronous robots with dynamic compasses, these sufficient conditions are also necessary.",
"We present a scalable robot motion planning algorithm for reach-avoid problems. We assume a discrete-time, linear model of the robot dynamics and a workspace described by a set of obstacles and a target region, where both the obstacles and the region are polyhedra. Our goal is to construct a trajectory, and the associated control strategy, that steers the robot from its initial point to the target while avoiding obstacles. Differently from previous approaches, based on the discretization of the continuous state space or uniform discretization of the workspace, our approach, inspired by the lazy satisfiability modulo theory paradigm, decomposes the planning problem into smaller subproblems, which can be efficiently solved using specialized solvers. At each iteration, we use a coarse, obstacle-based discretization of the workspace to obtain candidate high-level, discrete plans that solve a set of Boolean constraints, while completely abstracting the low-level continuous dynamics. The feasibility of the proposed plans is then checked via a convex program, under constraints on both the system dynamics and the control inputs, and new candidate plans are generated until a feasible one is found. To achieve scalability, we show how to generate succinct explanations for the infeasibility of a discrete plan by exploiting a relaxation of the convex program that allows detecting the earliest possible occurrence of an infeasible transition between workspace regions. Simulation results show that our algorithm favorably compares with state-of-the-art techniques and scales well for complex systems, including robot dynamics with up to 50 continuous states.",
"Consider a set of n finite set of simple autonomous mobile robots (asynchronous, no common coordinate system, no identities, no central coordination, no direct communication, no memory of the past, non-rigid, deterministic) initially in distinct locations, moving freely in the plane and able to sense the positions of the other robots. We study the primitive task of the robots arranging themselves on the vertices of a regular n-gon not fixed in advance (Uniform Circle Formation). In the literature, the existing algorithmic contributions are limited to conveniently restricted sets of initial configurations of the robots and to more powerful robots. The question of whether such simple robots could deterministically form a uniform circle has remained open. In this paper, we constructively prove that indeed the Uniform Circle Formation problem is solvable for any initial configuration in which the robots are in distinct locations, without any additional assumption (if two robots are in the same location, the problem is easily seen to be unsolvable). In addition to closing a long-standing problem, the result of this paper also implies that, for pattern formation, asynchrony is not a computational handicap, and that additional powers such as chirality and rigidity are computationally irrelevant.",
"We study verification problems for autonomous swarms of mobile robots that self-organize and cooperate to solve global objectives. In particular, we focus in this paper on the model proposed by Suzuki and Yamashita of anonymous robots evolving in a discrete space with a finite number of locations (here, a ring). A large number of algorithms have been proposed working for rings whose size is not a priori fixed and can be hence considered as a parameter. Handmade correctness proofs of these algorithms have been shown to be error-prone, and recent attention had been given to the application of formal methods to automatically prove those. Our work is the first to study the verification problem of such algorithms in the parameterized case. We show that safety and reachability problems are undecidable for robots evolving asynchronously. On the positive side, we show that safety properties are decidable in the synchronous case, as well as in the asynchronous case for a particular class of algorithms. Several properties on the protocol can be decided as well. Decision procedures rely on an encoding in Presburger arithmetics formulae that can be verified by an SMT-solver. Feasibility of our approach is demonstrated by the encoding of several case studies."
]
} |
1907.09871 | 2963864879 | The paper details the first successful attempt at using model-checking techniques to verify the correctness of distributed algorithms for robots evolving in a environment. The study focuses on the problem of rendezvous of two robots with lights. There exist many different rendezvous algorithms that aim at finding the minimal number of colors needed to solve rendezvous in various synchrony models (e.g., FSYNC, SSYNC, ASYNC). While these rendezvous algorithms are typically very simple, their analysis and proof of correctness tend to be extremely complex, tedious, and error-prone as impossibility results are based on subtle interactions between robots activation schedules. The paper presents a generic verification model written for the SPIN model-checker. In particular, we explain the subtle design decisions that allow to keep the search space finite and tractable, as well as prove several important theorems that support them. As a sanity check, we use the model to verify several known rendezvous algorithms in six different models of synchrony. In each case, we find that the results obtained from the model-checker are consistent with the results known in the literature. The model-checker outputs a counter-example execution in every case that is known to fail. In the course of developing and proving the validity of the model, we identified several fundamental theorems, including the ability for a well chosen algorithm and ASYNC scheduler to produce an emerging property of memory in a system of oblivious mobile robots, and why it is not a problem for luminous rendezvous algorithms. | On the other side, model checking and its derivatives (automatic program synthesis, parameterized model checking) hint at more automation once a suitable model has been defined with the input language of the model checker. In particular, model-checking proved useful to find bugs (usually in the setting) @cite_5 @cite_13 @cite_6 and to formally check the correctness of published algorithms @cite_32 @cite_5 @cite_28 . Automatic program synthesis @cite_11 @cite_1 was used to obtain automatically algorithms that are correct-by-design''. However, those approaches are limited to instances with few robots. Generalizing them to an arbitrary number of robots with similar models is doubtful as Sangnier al @cite_21 proved that safety and reachability problems are undecidable in the parameterized case. Another limitation of the above approaches is that they consider cases where mobile robots space (, graph). This limitation is due to the model used, that closely matches the original execution model by Suzuki and Yamashita @cite_19 . As a computer can only model a finite set of locations, a continuous 2D Euclidean space cannot be expressed in this model. | {
"cite_N": [
"@cite_28",
"@cite_21",
"@cite_1",
"@cite_32",
"@cite_6",
"@cite_19",
"@cite_5",
"@cite_13",
"@cite_11"
],
"mid": [
"2963688871",
"2097728881",
"2951766328",
"2001695893",
"2773498419",
"2562147820",
"2782960731",
"2101432564",
"1985861264"
],
"abstract": [
"We study verification problems for autonomous swarms of mobile robots that self-organize and cooperate to solve global objectives. In particular, we focus in this paper on the model proposed by Suzuki and Yamashita of anonymous robots evolving in a discrete space with a finite number of locations (here, a ring). A large number of algorithms have been proposed working for rings whose size is not a priori fixed and can be hence considered as a parameter. Handmade correctness proofs of these algorithms have been shown to be error-prone, and recent attention had been given to the application of formal methods to automatically prove those. Our work is the first to study the verification problem of such algorithms in the parameterized case. We show that safety and reachability problems are undecidable for robots evolving asynchronously. On the positive side, we show that safety properties are decidable in the synchronous case, as well as in the asynchronous case for a particular class of algorithms. Several properties on the protocol can be decided as well. Decision procedures rely on an encoding in Presburger arithmetics formulae that can be verified by an SMT-solver. Feasibility of our approach is demonstrated by the encoding of several case studies.",
"We address the pose mismatch problem which can occur in face verification systems that have only a single (frontal) face image available for training. In the framework of a Bayesian classifier based on mixtures of gaussians, the problem is tackled through extending each frontal face model with artificially synthesized models for non-frontal views. The synthesis methods are based on several implementations of maximum likelihood linear regression (MLLR), as well as standard multi-variate linear regression (LinReg). All synthesis techniques rely on prior information and learn how face models for the frontal view are related to face models for non-frontal views. The synthesis and extension approach is evaluated by applying it to two face verification systems: a holistic system (based on PCA-derived features) and a local feature system (based on DCT-derived features). Experiments on the FERET database suggest that for the holistic system, the LinReg-based technique is more suited than the MLLR-based techniques; for the local feature system, the results show that synthesis via a new MLLR implementation obtains better performance than synthesis based on traditional MLLR. The results further suggest that extending frontal models considerably reduces errors. It is also shown that the local feature system is less affected by view changes than the holistic system; this can be attributed to the parts based representation of the face, and, due to the classifier based on mixtures of gaussians, the lack of constraints on spatial relations between the face parts, allowing for deformations and movements of face areas.",
"The problem of automatically generating a computer program from some specification has been studied since the early days of AI. Recently, two competing approaches for automatic program learning have received significant attention: (1) neural program synthesis, where a neural network is conditioned on input output (I O) examples and learns to generate a program, and (2) neural program induction, where a neural network generates new outputs directly using a latent program representation. Here, for the first time, we directly compare both approaches on a large-scale, real-world learning task. We additionally contrast to rule-based program synthesis, which uses hand-crafted semantics to guide the program generation. Our neural models use a modified attention RNN to allow encoding of variable-sized sets of I O pairs. Our best synthesis model achieves 92 accuracy on a real-world test set, compared to the 34 accuracy of the previous best neural synthesis approach. The synthesis model also outperforms a comparable induction model on this task, but we more importantly demonstrate that the strength of each approach is highly dependent on the evaluation metric and end-user application. Finally, we show that we can train our neural models to remain very robust to the type of noise expected in real-world data (e.g., typos), while a highly-engineered rule-based system fails entirely.",
"This paper proposes a novel 3D scene interpretation approach for robots in mobile manipulation scenarios using a set of 3D point features (Fast Point Feature Histograms) and probabilistic graphical methods (Conditional Random Fields). Our system uses real time stereo with textured light to obtain dense depth maps in the robot's manipulators working space. For the purposes of manipulation, we want to interpret the planar supporting surfaces of the scene, recognize and segment the object classes into their primitive parts in 6 degrees of freedom (6DOF) so that the robot knows what it is attempting to use and where it may be handled. The scene interpretation algorithm uses a two-layer classification scheme: i) we estimate Fast Point Feature Histograms (FPFH) as local 3D point features to segment the objects of interest into geometric primitives; and ii) we learn and categorize object classes using a novel Global Fast Point Feature Histogram (GFPFH) scheme which uses the previously estimated primitives at each point. To show the validity of our approach, we analyze the proposed system for the problem of recognizing the object class of 20 objects in 500 table settings scenarios. Our algorithm identifies the planar surfaces, decomposes the scene and objects into geometric primitives with 98.27 accuracy and uses the geometric primitives to identify the object's class with an accuracy of 96.69 .",
"Probabilistic graphical models, such as partially observable Markov decision processes (POMDPs), have been used in stochastic spoken dialog systems to handle the inherent uncertainty in speech recognition and language understanding. Such dialog systems suffer from the fact that only a relatively small number of domain variables are allowed in the model, so as to ensure the generation of good-quality dialog policies. At the same time, the non-language perception modalities on robots, such as vision-based facial expression recognition and Lidar-based distance detection, can hardly be integrated into this process. In this paper, we use a probabilistic commonsense reasoner to “guide” our POMDP-based dialog manager, and present a principled, multimodal dialog management (MDM) framework that allows the robot's dialog belief state to be seamlessly updated by both observations of human spoken language, and exogenous events such as the change of human facial expressions. The MDM approach has been implemented and evaluated both in simulation and on a real mobile robot using guidance tasks.",
"We present a novel and scalable policy synthesis approach for robots. Rather than producing single-path plans for a static environment, we consider changing environments with uncontrollable agents, where the robot needs a policy to respond correctly over the infinite-horizon interaction with the environment. Our approach operates on task and motion domains, and combines actions over discrete states with continuous, collision-free paths. We synthesize a task and motion policy by iteratively generating a candidate policy and verifying its correctness. For efficient policy generation, we use grammars for potential policies to limit the search space and apply domain-specific heuristics to generalize verification failures, providing stricter constraints on policy candidates. For efficient policy verification, we construct compact, symbolic constraints for valid policies and employ a Satisfiability Modulo Theories (SMT) solver to check the validity of these constraints. Furthermore, the SMT solver enables quantitative specifications such as energy limits. The results show that our approach offers better scalability compared to a state-of-the-art policy synthesis tool in the tested benchmarks and demonstrate an order-of-magnitude speedup from our heuristics for the tested mobile manipulation domain.",
"Swarms of mobile robots recently attracted the focus of the Distributed Computing community. One of the fundamental problems in this context is that of exploration: the robots must coordinate to visit all locations that are reachable from their initial positions. Despite its apparent simplicity, this problem proved quite hard to characterise fully, due to many model variants, leading to informal error-prone reasoning. Over the past few years, a significant effort permitted to set up a formal framework, relying on the Coq proof assistant, which was used to provide certified results when robots evolve in a continuous bi-dimensional Euclidean space. However, the most challenging issues with exploration arise in the discrete setting (a.k.a. graph), where locations are modeled as vertices and where edges between vertices denote the ability for a robot to move from one location to the next. We present a formal model to tackle problems and reason about robot algorithms arising in the discrete setting. Our approach extends and generalises previous research efforts focusing on the continuous model. As case studies, we consider fundamental impossibility results for exploration with stop in the discrete model. To our knowledge, those are the first certified results in this context. This framework paves the way for a general certification workflow dedicated to mobile robots on graphs.",
"This paper describes a novel technique for the synthesis of imperative programs. Automated program synthesis has the potential to make programming and the design of systems easier by allowing programs to be specified at a higher-level than executable code. In our approach, which we call proof-theoretic synthesis, the user provides an input-output functional specification, a description of the atomic operations in the programming language, and a specification of the synthesized program's looping structure, allowed stack space, and bound on usage of certain operations. Our technique synthesizes a program, if there exists one, that meets the input-output specification and uses only the given resources. The insight behind our approach is to interpret program synthesis as generalized program verification, which allows us to bring verification tools and techniques to program synthesis. Our synthesis algorithm works by creating a program with unknown statements, guards, inductive invariants, and ranking functions. It then generates constraints that relate the unknowns and enforces three kinds of requirements: partial correctness, loop termination, and well-formedness conditions on program guards. We formalize the requirements that program verification tools must meet to solve these constraint and use tools from prior work as our synthesizers. We demonstrate the feasibility of the proposed approach by synthesizing programs in three different domains: arithmetic, sorting, and dynamic programming. Using verification tools that we previously built in the VS3 project we are able to synthesize programs for complicated arithmetic algorithms including Strassen's matrix multiplication and Bresenham's line drawing; several sorting algorithms; and several dynamic programming algorithms. For these programs, the median time for synthesis is 14 seconds, and the ratio of synthesis to verification time ranges between 1x to 92x (with an median of 7x), illustrating the potential of the approach.",
"Model checking and supervisor synthesis have been successful in solving different design problems related to discrete systems in the last decades. In this paper, we analyze some advantages and drawbacks of these approaches and combine them for mutual improvement. We achieve this through a generalization of the supervisory control problem proposed by Ramadge and Wonham. The objective of that problem is to synthesize a supervisor which constrains a system's behavior according to a given specification, ensuring controllability and coaccessibility. By introducing a new representation of the solution using systems of μ-calculus equations, we are able to handle these two conditions separately and thus to exchange the coaccessibility requirement by any condition that could be used in model checking. Well-known results on μ-calculus model checking allow us to easily assess the computational complexity of any generalization. Moreover, the model checking approach also delivers algorithms to solve the generalized synthesis problem. We include an example in which the coaccessibility requirement is replaced by fairness constraints. The paper also contains an analysis of related work by several authors."
]
} |
1907.09939 | 2963333474 | Understanding large amounts of spatiotemporal data from particle-based simulations, such as molecular dynamics, often relies on the computation and analysis of aggregate measures. These, however, by virtue of aggregation, hide structural information about the space time localization of the studied phenomena. This leads to degenerate cases where the measures fail to capture distinct behaviour. In order to drill into these aggregate values, we propose a multi-scale visual exploration technique. Our novel representation, based on partial domain aggregation, enables the construction of a continuous scale-space for discrete datasets and the simultaneous exploration of scales in both space and time. We link these two scale-spaces in a scale-space space-time cube and model linked views as orthogonal slices through this cube, thus enabling the rapid identification of spatio-temporal patterns at multiple scales. To demonstrate the effectiveness of our approach, we showcase an advanced exploration of a protein-ligand simulation. | This paper contributes to the analysis of data that usually would be treated as a time series of spatially aggregated values. A good overview of visualization techniques for time-dependent data can be found in a book by @cite_57 . They focus primarily on time-dependent data in general, without specifically addressing spatial localization or scale-space approaches. @cite_28 proposed an analysis based on multivariate trend identification in time dependent data without spatial dependence. @cite_9 explore the use of parallel coordinates for the visualization of trajectories in multi-dimensional dynamical systems. | {
"cite_N": [
"@cite_57",
"@cite_9",
"@cite_28"
],
"mid": [
"2020141201",
"2141859737",
"1925915230"
],
"abstract": [
"We introduce a novel projection-based visualization method for high-dimensional data sets by combining concepts from MDS and the geometry of the hyperbolic spaces. This approach hyperbolic multi-dimensional scaling (H-MDS) is a synthesis of two important concepts for explorative data analysis and visualization: (i) multi-dimensional scaling uses proximity or pair distance data to generate a low-dimensional, spatial presentation of the data; (ii) previous work on the \"hyperbolic tree browser\" demonstrated the extraordinary advantages for an interactive display of graph-like data in the two-dimensional hyperbolic space (H2).In the new approach, H-MDS maps proximity data directly into the H2. This removes the restriction to \"quasihierarchical\", graph-based data--a major limitation of (ii). Since a suitable distance function can convert all kinds of data to proximity (or distance-based) data, this type of data can be considered the most general.We review important properties of the hyperbolic space and, in particular, the circular Poincare model of the H2. It enables effective human-computer interaction: by mouse dragging the \"focus\", the user can navigate in the data without loosing the context. In H2 the \"fish-eye\" behavior originates not simply by a non-linear view transformation but rather by extraordinary, non-Euclidean properties of the H2. Especially, the exponential growth of length and area of the underlying space makes the H2 a prime target for mapping hierarchical and (now also) high-dimensional data.Several high-dimensional mapping examples including synthetic and real-world data are presented. Since high-dimensional data produce \"ring\"-shaped displays, we present methods to enhance the display by modulating the dissimilarity contrast. This is demonstrated for an application for unstructured text: i.e., by using multiple film critiques from news:rec.art.movies.reviews and www.imdb.com, each movie is placed within the H2--creating a \"space of movies\" for interactive exploration.",
"This paper is concerned with the representation and recognition of the observed dynamics (i.e., excluding purely spatial appearance cues) of spacetime texture based on a spatiotemporal orientation analysis. The term “spacetime texture” is taken to refer to patterns in visual spacetime, (x,y,t), that primarily are characterized by the aggregate dynamic properties of elements or local measurements accumulated over a region of spatiotemporal support, rather than in terms of the dynamics of individual constituents. Examples include image sequences of natural processes that exhibit stochastic dynamics (e.g., fire, water, and windblown vegetation) as well as images of simpler dynamics when analyzed in terms of aggregate region properties (e.g., uniform motion of elements in imagery, such as pedestrians and vehicular traffic). Spacetime texture representation and recognition is important as it provides an early means of capturing the structure of an ensuing image stream in a meaningful fashion. Toward such ends, a novel approach to spacetime texture representation and an associated recognition method are described based on distributions (histograms) of spacetime orientation structure. Empirical evaluation on both standard and original image data sets shows the promise of the approach, including significant improvement over alternative state-of-the-art approaches in recognizing the same pattern from different viewpoints.",
"Many scientific and economic problems involve the analysis of highdimensional time series datasets. However, theoretical studies in highdimensional statistics to date rely primarily on the assumption of independent and identically distributed (i.i.d.) samples. In this work, we focus on stable Gaussian processes and investigate the theoretical properties of � 1-regularized estimates in two important statistical problems in the context of high-dimensional time series: (a) stochastic regression with serially correlated errors and (b) transition matrix estimation in vector autoregressive (VAR) models. We derive nonasymptotic upper bounds on the estimation errors of the regularized estimates and establish that consistent estimation under high-dimensional scaling is possible via � 1-regularization for a large class of stable processes under sparsity constraints. A key technical contribution of the work is to introduce a measure of stability for stationary processes using their spectral properties that provides insight into the effect of dependence on the accuracy of the regularized estimates. With this proposed stability measure, we establish some useful deviation bounds for dependent data, which can be used to study several important regularized estimates in a time series setting."
]
} |
1907.09939 | 2963333474 | Understanding large amounts of spatiotemporal data from particle-based simulations, such as molecular dynamics, often relies on the computation and analysis of aggregate measures. These, however, by virtue of aggregation, hide structural information about the space time localization of the studied phenomena. This leads to degenerate cases where the measures fail to capture distinct behaviour. In order to drill into these aggregate values, we propose a multi-scale visual exploration technique. Our novel representation, based on partial domain aggregation, enables the construction of a continuous scale-space for discrete datasets and the simultaneous exploration of scales in both space and time. We link these two scale-spaces in a scale-space space-time cube and model linked views as orthogonal slices through this cube, thus enabling the rapid identification of spatio-temporal patterns at multiple scales. To demonstrate the effectiveness of our approach, we showcase an advanced exploration of a protein-ligand simulation. | Our approach is based on the concept of the space-time cube, coined by H "a gerstrand @cite_16 in 1970. Since then it was repeatedly used, either explicitly or as an underlying concept. Recently, @cite_35 presented a useful overview of related techniques. They describe the theoretical concept of a generalized space-time cube together with a taxonomy of all elementary space-time operations and their combinations that can be performed on such a space-time cube. A more abstract approach to time-dependent volume data analysis has been proposed by @cite_41 , based on hyperplane slicing of a four-dimensional space-time hypercube. Other examples of slicing higher-dimensional data (not necessarily time dependent) include Sliceplorer @cite_0 , Hyperslice @cite_19 , Hypersliceplorer @cite_26 and HyperMoVal @cite_27 . | {
"cite_N": [
"@cite_35",
"@cite_26",
"@cite_41",
"@cite_0",
"@cite_19",
"@cite_27",
"@cite_16"
],
"mid": [
"2393302563",
"2041855012",
"2097621063",
"2010548775",
"2461259506",
"2508919161",
"2020141201"
],
"abstract": [
"We present the generalized space-time cube, a descriptive model for visualizations of temporal data. Visualizations are described as operations on the cube, which transform the cube's 3D shape into readable 2D visualizations. Operations include extracting subparts of the cube, flattening it across space or time or transforming the cubes geometry and content. We introduce a taxonomy of elementary space-time cube operations and explain how these operations can be combined and parameterized. The generalized space-time cube has two properties: 1 it is purely conceptual without the need to be implemented, and 2 it applies to all datasets that can be represented in two dimensions plus time e.g. geo-spatial, videos, networks, multivariate data. The proper choice of space-time cube operations depends on many factors, for example, density or sparsity of a cube. Hence, we propose a characterization of structures within space-time cubes, which allows us to discuss strengths and limitations of operations. We finally review interactive systems that support multiple operations, allowing a user to customize his view on the data. With this framework, we hope to facilitate the description, criticism and comparison of temporal data visualizations, as well as encourage the exploration of new techniques and systems. This paper is an extension ofBach etal.'s 2014 work.",
"We introduce a volumetric space-time technique for the reconstruction of moving and deforming objects from point data. The output of our method is a four-dimensional space-time solid, made up of spatial slices, each of which is a three-dimensional solid bounded by a watertight manifold. The motion of the object is described as an incompressible flow of material through time. We optimize the flow so that the distance material moves from one time frame to the next is bounded, the density of material remains constant, and the object remains compact. This formulation overcomes deficiencies in the acquired data, such as persistent occlusions, errors, and missing frames. We demonstrate the performance of our flow-based technique by reconstructing coherent sequences of watertight models from incomplete scanner data.",
"We present an alternative method for viewing time-varying volumetric data. We consider such data as a four-dimensional data field, rather than considering space and time as separate entities. If we treat the data in this manner, we can apply high dimensional slicing and projection techniques to generate an image hyperplane. The user is provided with an intuitive user interface to specify arbitrary hyperplanes in 4D, which can be displayed with standard volume rendering techniques. From the volume specification, we are able to extract arbitrary hyperslices, combine slices together into a hyperprojection volume, or apply a 4D raycasting method to generate the same results. In combination with appropriate integration operators and transfer functions, we are able to extract and present different space-time features to the user.",
"This paper describes a generalized axiomatic scale-space theory that makes it possible to derive the notions of linear scale-space, affine Gaussian scale-space and linear spatio-temporal scale-space using a similar set of assumptions (scale-space axioms). The notion of non-enhancement of local extrema is generalized from previous application over discrete and rotationally symmetric kernels to continuous and more general non-isotropic kernels over both spatial and spatio-temporal image domains. It is shown how a complete classification can be given of the linear (Gaussian) scale-space concepts that satisfy these conditions on isotropic spatial, non-isotropic spatial and spatio-temporal domains, which results in a general taxonomy of Gaussian scale-spaces for continuous image data. The resulting theory allows filter shapes to be tuned from specific context information and provides a theoretical foundation for the recently exploited mechanisms of shape adaptation and velocity adaptation, with highly useful applications in computer vision. It is also shown how time-causal spatio-temporal scale-spaces can be derived from similar assumptions. The mathematical structure of these scale-spaces is analyzed in detail concerning transformation properties over space and time, the temporal cascade structure they satisfy over time as well as properties of the resulting multi-scale spatio-temporal derivative operators. It is also shown how temporal derivatives with respect to transformed time can be defined, leading to the formulation of a novel analogue of scale normalized derivatives for time-causal scale-spaces. The kernels generated from these two types of theories have interesting relations to biological vision. We show how filter kernels generated from the Gaussian spatio-temporal scale-space as well as the time-causal spatio-temporal scale-space relate to spatio-temporal receptive field profiles registered from mammalian vision. Specifically, we show that there are close analogies to space-time separable cells in the LGN as well as to both space-time separable and non-separable cells in the striate cortex. We do also present a set of plausible models for complex cells using extended quasi-quadrature measures expressed in terms of scale normalized spatio-temporal derivatives. The theories presented as well as their relations to biological vision show that it is possible to describe a general set of Gaussian and or time-causal scale-spaces using a unified framework, which generalizes and complements previously presented scale-space formulations in this area.",
"Time parallel time integration methods have received renewed interest over the last decade because of the advent of massively parallel computers, which is mainly due to the clock speed limit reached on today’s processors. When solving time dependent partial differential equations, the time direction is usually not used for parallelization. But when parallelization in space saturates, the time direction offers itself as a further direction for parallelization. The time direction is however special, and for evolution problems there is a causality principle: the solution later in time is affected (it is even determined) by the solution earlier in time, but not the other way round. Algorithms trying to use the time direction for parallelization must therefore be special, and take this very different property of the time dimension into account.We show in this chapter how time domain decomposition methods were invented, and give an overview of the existing techniques. Time parallel methods can be classified into four different groups: methods based on multiple shooting, methods based on domain decomposition and waveform relaxation, space-time multigrid methods and direct time parallel methods. We show for each of these techniques the main inventions over time by choosing specific publications and explaining the core ideas of the authors. This chapter is for people who want to quickly gain an overview of the exciting and rapidly developing area of research of time parallel methods.",
"[See the paper for the full abstract.] We show tight upper and lower bounds for time-space trade-offs for the @math -Approximate Near Neighbor Search problem. For the @math -dimensional Euclidean space and @math -point datasets, we develop a data structure with space @math and query time @math for every @math such that: This is the first data structure that achieves sublinear query time and near-linear space for every approximation factor @math , improving upon [Kapralov, PODS 2015]. The data structure is a culmination of a long line of work on the problem for all space regimes; it builds on Spherical Locality-Sensitive Filtering [Becker, Ducas, Gama, Laarhoven, SODA 2016] and data-dependent hashing [Andoni, Indyk, Nguyen, Razenshteyn, SODA 2014] [Andoni, Razenshteyn, STOC 2015]. Our matching lower bounds are of two types: conditional and unconditional. First, we prove tightness of the whole above trade-off in a restricted model of computation, which captures all known hashing-based approaches. We then show unconditional cell-probe lower bounds for one and two probes that match the above trade-off for @math , improving upon the best known lower bounds from [Panigrahy, Talwar, Wieder, FOCS 2010]. In particular, this is the first space lower bound (for any static data structure) for two probes which is not polynomially smaller than the one-probe bound. To show the result for two probes, we establish and exploit a connection to locally-decodable codes.",
"We introduce a novel projection-based visualization method for high-dimensional data sets by combining concepts from MDS and the geometry of the hyperbolic spaces. This approach hyperbolic multi-dimensional scaling (H-MDS) is a synthesis of two important concepts for explorative data analysis and visualization: (i) multi-dimensional scaling uses proximity or pair distance data to generate a low-dimensional, spatial presentation of the data; (ii) previous work on the \"hyperbolic tree browser\" demonstrated the extraordinary advantages for an interactive display of graph-like data in the two-dimensional hyperbolic space (H2).In the new approach, H-MDS maps proximity data directly into the H2. This removes the restriction to \"quasihierarchical\", graph-based data--a major limitation of (ii). Since a suitable distance function can convert all kinds of data to proximity (or distance-based) data, this type of data can be considered the most general.We review important properties of the hyperbolic space and, in particular, the circular Poincare model of the H2. It enables effective human-computer interaction: by mouse dragging the \"focus\", the user can navigate in the data without loosing the context. In H2 the \"fish-eye\" behavior originates not simply by a non-linear view transformation but rather by extraordinary, non-Euclidean properties of the H2. Especially, the exponential growth of length and area of the underlying space makes the H2 a prime target for mapping hierarchical and (now also) high-dimensional data.Several high-dimensional mapping examples including synthetic and real-world data are presented. Since high-dimensional data produce \"ring\"-shaped displays, we present methods to enhance the display by modulating the dissimilarity contrast. This is demonstrated for an application for unstructured text: i.e., by using multiple film critiques from news:rec.art.movies.reviews and www.imdb.com, each movie is placed within the H2--creating a \"space of movies\" for interactive exploration."
]
} |
1907.09939 | 2963333474 | Understanding large amounts of spatiotemporal data from particle-based simulations, such as molecular dynamics, often relies on the computation and analysis of aggregate measures. These, however, by virtue of aggregation, hide structural information about the space time localization of the studied phenomena. This leads to degenerate cases where the measures fail to capture distinct behaviour. In order to drill into these aggregate values, we propose a multi-scale visual exploration technique. Our novel representation, based on partial domain aggregation, enables the construction of a continuous scale-space for discrete datasets and the simultaneous exploration of scales in both space and time. We link these two scale-spaces in a scale-space space-time cube and model linked views as orthogonal slices through this cube, thus enabling the rapid identification of spatio-temporal patterns at multiple scales. To demonstrate the effectiveness of our approach, we showcase an advanced exploration of a protein-ligand simulation. | Space reformation techniques are valuable tools for gaining insight into the spatial structure of the data. Recent publications present methods for a decomposition using a space-filling curve in MotionRugs @cite_53 and Dynamic Volume Lines @cite_1 . These works transform continuous volume data into one-dimensional representations by following a space-filling curve through the volume. As this kind of space reformation suffers from delocalization , where points close to each other can end up far apart in the representation, @cite_30 tried to overcome this limitation by context-based space-filling curves. A space-filling curve is not applicable to our case as it does not provide the desired aggregation required for the statistical treatment of particle data. Several works address space reformation techniques for volume data, of which we highlight two: A general approach to volume transformation by @cite_6 , via the use of spatial transfer functions for 3D volume warping, and a curved planar reformation by @cite_17 , that was used in medical visualization. For more examples, @cite_38 provide a survey of flattening based techniques in medical visualization. | {
"cite_N": [
"@cite_30",
"@cite_38",
"@cite_53",
"@cite_1",
"@cite_6",
"@cite_17"
],
"mid": [
"2892217449",
"2003653728",
"1994514622",
"2024251722",
"2609754928",
"2949130266"
],
"abstract": [
"Despite remarkable advances in image synthesis research, existing works often fail in manipulating images under the context of large geometric transformations. Synthesizing person images conditioned on arbitrary poses is one of the most representative examples where the generation quality largely relies on the capability of identifying and modeling arbitrary transformations on different body parts. Current generative models are often built on local convolutions and overlook the key challenges (e.g. heavy occlusions, different views or dramatic appearance changes) when distinct geometric changes happen for each part, caused by arbitrary pose manipulations. This paper aims to resolve these challenges induced by geometric variability and spatial displacements via a new Soft-Gated Warping Generative Adversarial Network (Warping-GAN), which is composed of two stages: 1) it first synthesizes a target part segmentation map given a target pose, which depicts the region-level spatial layouts for guiding image synthesis with higher-level structure constraints; 2) the Warping-GAN equipped with a soft-gated warping-block learns feature-level mapping to render textures from the original image into the generated segmentation map. Warping-GAN is capable of controlling different transformation degrees given distinct target poses. Moreover, the proposed warping-block is light-weight and flexible enough to be injected into any networks. Human perceptual studies and quantitative evaluations demonstrate the superiority of our Warping-GAN that significantly outperforms all existing methods on two large datasets.",
"Space-filling visualization techniques have proved their capability in visualizing large hierarchical structured data. However, most existing techniques restrict their partitioning process in vertical and horizontal direction only, which cause problem with identifying hierarchical structures. This paper presents a new space-filling method named Angular Treemaps that relax the constraint of the rectangular subdivision. The approach of Angular Treemaps utilizes divide and conquer paradigm to visualize and emphasize large hierarchical structures within a compact and limited display area with better interpretability. Angular Treemaps generate various layouts to highlight hierarchical sub-structure based on user's preferences or system recommendations. It offers flexibility to be adopted into a wider range of applications, regarding different enclosing shapes. Preliminary usability results suggest user's performance by using this technique is improved in locating and identifying categorized analysis tasks.",
"Capturing an enclosing volume of moving subjects and organs using fast individual image slice acquisition has shown promise in dealing with motion artefacts. Motion between slice acquisitions results in spatial inconsistencies that can be resolved by slice-to-volume reconstruction (SVR) methods to provide high quality 3D image data. Existing algorithms are, however, typically very slow, specialised to specific applications and rely on approximations, which impedes their potential clinical use. In this paper, we present a fast multi-GPU accelerated framework for slice-to-volume reconstruction. It is based on optimised 2D 3D registration, super-resolution with automatic outlier rejection and an additional (optional) intensity bias correction. We introduce a novel and fully automatic procedure for selecting the image stack with least motion to serve as an initial registration target. We evaluate the proposed method using artificial motion corrupted phantom data as well as clinical data, including tracked freehand ultrasound of the liver and fet al Magnetic Resonance Imaging. We achieve speed-up factors greater than 30 compared to a single CPU system and greater than 10 compared to currently available state-of-the-art multi-core CPU methods. We ensure high reconstruction accuracy by exact computation of the point-spread function for every input data point, which has not previously been possible due to computational limitations. Our framework and its implementation is scalable for available computational infrastructures and tests show a speed-up factor of 1.70 for each additional GPU. This paves the way for the online application of image based reconstruction methods during clinical examinations. The source code for the proposed approach is publicly available.",
"We propose a method for the reconstruction of volumetric fet al MRI from 2D slices, comprising super-resolution reconstruction of the volume interleaved with slice-to-volume registration to correct for the motion. The method incorporates novel intensity matching of acquired 2D slices and robust statistics which completely excludes identified misregistered or corrupted voxels and slices. The reconstruction method is applied to motion-corrupted data simulated from MRI of a preterm neonate, as well as 10 clinically acquired thick-slice fet al MRI scans and three scan-sequence optimized thin-slice fet al datasets. The proposed method produced high quality reconstruction results from all the datasets to which it was applied. Quantitative analysis performed on simulated and clinical data shows that both intensity matching and robust statistics result in statistically significant improvement of super-resolution reconstruction. The proposed novel EM-based robust statistics also improves the reconstruction when compared to previously proposed Huber robust statistics. The best results are obtained when thin-slice data and the correct approximation of the point spread function is used. This paper addresses the need for a comprehensive reconstruction algorithm of 3D fet al MRI, so far lacking in the scientific literature.",
"In this paper, we present a learning based approach to depth fusion, i.e., dense 3D reconstruction from multiple depth images. The most common approach to depth fusion is based on averaging truncated signed distance functions, which was originally proposed by Curless and Levoy in 1996. While this method is simple and provides great results, it is not able to reconstruct (partially) occluded surfaces and requires a large number frames to filter out sensor noise and outliers. Motivated by the availability of large 3D model repositories and recent advances in deep learning, we present a novel 3D CNN architecture that learns to predict an implicit surface representation from the input depth maps. Our learning based method significantly outperforms the traditional volumetric fusion approach in terms of noise reduction and outlier suppression. By learning the structure of real world 3D objects and scenes, our approach is further able to reconstruct occluded regions and to fill in gaps in the reconstruction. We demonstrate that our learning based approach outperforms both vanilla TSDF fusion as well as TV-L1 fusion on the task of volumetric fusion. Further, we demonstrate state-of-the-art 3D shape completion results.",
"In this paper, we present a learning based approach to depth fusion, i.e., dense 3D reconstruction from multiple depth images. The most common approach to depth fusion is based on averaging truncated signed distance functions, which was originally proposed by Curless and Levoy in 1996. While this method is simple and provides great results, it is not able to reconstruct (partially) occluded surfaces and requires a large number frames to filter out sensor noise and outliers. Motivated by the availability of large 3D model repositories and recent advances in deep learning, we present a novel 3D CNN architecture that learns to predict an implicit surface representation from the input depth maps. Our learning based method significantly outperforms the traditional volumetric fusion approach in terms of noise reduction and outlier suppression. By learning the structure of real world 3D objects and scenes, our approach is further able to reconstruct occluded regions and to fill in gaps in the reconstruction. We demonstrate that our learning based approach outperforms both vanilla TSDF fusion as well as TV-L1 fusion on the task of volumetric fusion. Further, we demonstrate state-of-the-art 3D shape completion results."
]
} |
1907.09939 | 2963333474 | Understanding large amounts of spatiotemporal data from particle-based simulations, such as molecular dynamics, often relies on the computation and analysis of aggregate measures. These, however, by virtue of aggregation, hide structural information about the space time localization of the studied phenomena. This leads to degenerate cases where the measures fail to capture distinct behaviour. In order to drill into these aggregate values, we propose a multi-scale visual exploration technique. Our novel representation, based on partial domain aggregation, enables the construction of a continuous scale-space for discrete datasets and the simultaneous exploration of scales in both space and time. We link these two scale-spaces in a scale-space space-time cube and model linked views as orthogonal slices through this cube, thus enabling the rapid identification of spatio-temporal patterns at multiple scales. To demonstrate the effectiveness of our approach, we showcase an advanced exploration of a protein-ligand simulation. | Substantial work has been done on the visualization of trajectory-based simulation data, including projects such as OVITO @cite_34 , Trillion Particles @cite_23 , Multiscale HIV @cite_13 . These are primarily concerned with the sheer volume of the data and its direct visualization, rather than the computation of derived properties and their temporal analysis. | {
"cite_N": [
"@cite_34",
"@cite_13",
"@cite_23"
],
"mid": [
"2125184461",
"2482726005",
"2020141201"
],
"abstract": [
"Petascale plasma physics simulations have recently entered the regime of simulating trillions of particles. These unprecedented simulations generate massive amounts of data, posing significant challenges in storage, analysis, and visualization. In this paper, we present parallel I O, analysis, and visualization results from a VPIC trillion particle simulation running on 120,000 cores, which produces 30TB of data for a single timestep. We demonstrate the successful application of H5Part, a particle data extension of parallel HDF5, for writing the dataset at a significant fraction of system peak I O rates. To enable efficient analysis, we develop hybrid parallel FastQuery to index and query data using multi-core CPUs on distributed memory hardware. We show good scalability results for the FastQuery implementation using up to 10,000 cores. Finally, we apply this indexing query-driven approach to facilitate the first-ever analysis and visualization of the trillion particle dataset.",
"Current approaches for visual--inertial odometry (VIO) are able to attain highly accurate state estimation via nonlinear optimization. However, real-time optimization quickly becomes infeasible as the trajectory grows over time; this problem is further emphasized by the fact that inertial measurements come at high rate, hence, leading to the fast growth of the number of variables in the optimization. In this paper, we address this issue by preintegrating inertial measurements between selected keyframes into single relative motion constraints. Our first contribution is a preintegration theory that properly addresses the manifold structure of the rotation group. We formally discuss the generative measurement model as well as the nature of the rotation noise and derive the expression for the maximum a posteriori state estimator. Our theoretical development enables the computation of all necessary Jacobians for the optimization and a posteriori bias correction in analytic form. The second contribution is to show that the preintegrated inertial measurement unit model can be seamlessly integrated into a visual--inertial pipeline under the unifying framework of factor graphs. This enables the application of incremental-smoothing algorithms and the use of a structureless model for visual measurements, which avoids optimizing over the 3-D points, further accelerating the computation. We perform an extensive evaluation of our monocular VIO pipeline on real and simulated datasets. The results confirm that our modeling effort leads to an accurate state estimation in real time, outperforming state-of-the-art approaches.",
"We introduce a novel projection-based visualization method for high-dimensional data sets by combining concepts from MDS and the geometry of the hyperbolic spaces. This approach hyperbolic multi-dimensional scaling (H-MDS) is a synthesis of two important concepts for explorative data analysis and visualization: (i) multi-dimensional scaling uses proximity or pair distance data to generate a low-dimensional, spatial presentation of the data; (ii) previous work on the \"hyperbolic tree browser\" demonstrated the extraordinary advantages for an interactive display of graph-like data in the two-dimensional hyperbolic space (H2).In the new approach, H-MDS maps proximity data directly into the H2. This removes the restriction to \"quasihierarchical\", graph-based data--a major limitation of (ii). Since a suitable distance function can convert all kinds of data to proximity (or distance-based) data, this type of data can be considered the most general.We review important properties of the hyperbolic space and, in particular, the circular Poincare model of the H2. It enables effective human-computer interaction: by mouse dragging the \"focus\", the user can navigate in the data without loosing the context. In H2 the \"fish-eye\" behavior originates not simply by a non-linear view transformation but rather by extraordinary, non-Euclidean properties of the H2. Especially, the exponential growth of length and area of the underlying space makes the H2 a prime target for mapping hierarchical and (now also) high-dimensional data.Several high-dimensional mapping examples including synthetic and real-world data are presented. Since high-dimensional data produce \"ring\"-shaped displays, we present methods to enhance the display by modulating the dissimilarity contrast. This is demonstrated for an application for unstructured text: i.e., by using multiple film critiques from news:rec.art.movies.reviews and www.imdb.com, each movie is placed within the H2--creating a \"space of movies\" for interactive exploration."
]
} |
1907.09939 | 2963333474 | Understanding large amounts of spatiotemporal data from particle-based simulations, such as molecular dynamics, often relies on the computation and analysis of aggregate measures. These, however, by virtue of aggregation, hide structural information about the space time localization of the studied phenomena. This leads to degenerate cases where the measures fail to capture distinct behaviour. In order to drill into these aggregate values, we propose a multi-scale visual exploration technique. Our novel representation, based on partial domain aggregation, enables the construction of a continuous scale-space for discrete datasets and the simultaneous exploration of scales in both space and time. We link these two scale-spaces in a scale-space space-time cube and model linked views as orthogonal slices through this cube, thus enabling the rapid identification of spatio-temporal patterns at multiple scales. To demonstrate the effectiveness of our approach, we showcase an advanced exploration of a protein-ligand simulation. | Focusing on the trajectories, hierarchical particle grouping for large datasets has been done by @cite_33 . @cite_43 extract prominent trajectories from large particle data. present a specialized tool for exploring Monte Carlo simulations of photo-voltaic cells. | {
"cite_N": [
"@cite_43",
"@cite_33"
],
"mid": [
"1967509282",
"2081455211"
],
"abstract": [
"Interactive visualization of large particle sets is required to analyze the complicated structures and formation processes in astrophysical particle simulations. While some research has been done on the development of visualization techniques for steady particle fields, only very few approaches have been proposed to interactively visualize large time-varying fields and their dynamics. Particle trajectories are known to visualize dynamic processes over time, but due to occlusion and visual cluttering such techniques have only been reported for very small particle sets so far. In this paper we present a novel technique to solve these problems, and we demonstrate the potential of our approach for the visual exploration of large astrophysical particle sequences. We present a new hierarchical space-time data structure for particle sets which allows for a scale-space analysis of trajectories in the simulated fields. In combination with visualization techniques that adapt to the respective scales, clusters of particles with homogeneous motion as well as separation and merging regions can be identified effectively. The additional use of mapping functions to modulate the color and size of trajectories allows emphasizing various particle properties like direction, speed, or particle-specific attributes like temperature. Furthermore, tracking of interactively selected particle subsets permits the user to focus on structures of interest.",
"Methods to extract information from the tracking of mobile objects particles have broad interest in biological and physical sciences. Techniques based on simple criteria of proximity in time-consecutive snapshots are useful to identify the trajectories of the particles. However, they become problematic as the motility and or the density of the particles increases due to uncertainties on the trajectories that particles followed during the images’ acquisition time. Here, we report an efficient method for learning parameters of the dynamics of the particles from their positions in time-consecutive images. Our algorithm belongs to the class of message-passing algorithms, known in computer science, information theory, and statistical physics as belief propagation (BP). The algorithm is distributed, thus allowing parallel implementation suitable for computations on multiple machines without significant intermachine overhead. We test our method on the model example of particle tracking in turbulent flows, which is particularly challenging due to the strong transport that those flows produce. Our numerical experiments show that the BP algorithm compares in quality with exact Markov Chain Monte Carlo algorithms, yet BP is far superior in speed. We also suggest and analyze a random distance model that provides theoretical justification for BP accuracy. Methods developed here systematically formulate the problem of particle tracking and provide fast and reliable tools for the model’s extensive range of applications."
]
} |
1907.09939 | 2963333474 | Understanding large amounts of spatiotemporal data from particle-based simulations, such as molecular dynamics, often relies on the computation and analysis of aggregate measures. These, however, by virtue of aggregation, hide structural information about the space time localization of the studied phenomena. This leads to degenerate cases where the measures fail to capture distinct behaviour. In order to drill into these aggregate values, we propose a multi-scale visual exploration technique. Our novel representation, based on partial domain aggregation, enables the construction of a continuous scale-space for discrete datasets and the simultaneous exploration of scales in both space and time. We link these two scale-spaces in a scale-space space-time cube and model linked views as orthogonal slices through this cube, thus enabling the rapid identification of spatio-temporal patterns at multiple scales. To demonstrate the effectiveness of our approach, we showcase an advanced exploration of a protein-ligand simulation. | @cite_47 present a visual analytics tool for exploring molecular structures based on the Solvent Accessible Surface (SAS). This is a geometric approach to the extraction of the spatial configuration of a protein--ligand interaction. MoleCollar and Tunnel Heat Map @cite_24 are works by By s where they reform the properties of a protein tunnel into the tunnel's centre line and the tunnel's cross section. In AnimoAminoMiner @cite_31 , the authors focus on amino acids lining the tunnel and their temporal development. | {
"cite_N": [
"@cite_24",
"@cite_47",
"@cite_31"
],
"mid": [
"2160715766",
"1879416943",
"2027638447"
],
"abstract": [
"In this paper we propose a novel method for the interactive exploration of protein tunnels. The basic principle of our approach is that we entirely abstract from the 3D 4D space the simulated phenomenon is embedded in. A complex 3D structure and its curvature information is represented only by a straightened tunnel centerline and its width profile. This representation focuses on a key aspect of the studied geometry and frees up graphical estate to key chemical and physical properties represented by surrounding amino acids. The method shows the detailed tunnel profile and its temporal aggregation. The profile is interactively linked with a visual overview of all amino acids which are lining the tunnel over time. In this overview, each amino acid is represented by a set of colored lines depicting the spatial and temporal impact of the amino acid on the corresponding tunnel. This representation clearly shows the importance of amino acids with respect to selected criteria. It helps the biochemists to select the candidate amino acids for mutation which changes the protein function in a desired way. The AnimoAminoMiner was designed in close cooperation with domain experts. Its usefulness is documented by their feedback and a case study, which are included.",
"Studying the characteristics of proteins and their inner void space, including their geometry, physico-chemical properties and dynamics are instrumental for evaluating the reactivity of the protein with other small molecules. The analysis of long simulations of molecular dynamics produces a large number of voids which have to be further explored and evaluated. In this paper we propose three new methods: two of them convey important properties along the long axis of a selected void during molecular dynamics and one provides a comprehensive picture across the void. The first two proposed methods use a specific heat map to present two types of information: an overview of all detected tunnels in the dynamics and their bottleneck width and stability over time, and an overview of a specific tunnel in the dynamics showing the bottleneck position and changes of the tunnel length over time. These methods help to select a small subset of tunnels, which are explored individually and in detail. For this stage we propose the third method, which shows in one static image the temporal evolvement of the shape of the most critical tunnel part, i.e., its bottleneck. This view is enriched with abstract depictions of different physico-chemical properties of the amino acids surrounding the bottleneck. The usefulness of our newly proposed methods is demonstrated on a case study and the feedback from the domain experts is included. The biochemists confirmed that our novel methods help to convey the information about the appearance and properties of tunnels in a very intuitive and comprehensible manner.",
"Abstract Two major components are required for a successful prediction of the three-dimensional structure of peptides and proteins: an efficient global optimization procedure which is capable of finding an appropriate local minimum for the strongly anisotropic function of hundreds of variables, and a set of free energy components for a protein molecule in solution which are computationally inexpensive enough to be used in the search procedure, yet sufficiently accurate to ensure the uniqueness of the native conformation. We here found an efficient way to make a random step in a Monte Carlo procedure given knowledge of the energy or statistical properties of conformational subspaces (e.g. φ-ψ zones or side-chain torsion angles). This biased probability Monte Carlo (BPMC) procedure randomly selects the subspace first, then makes a step to a new random position independent of the previous position, but according to the predefined continuous probability distribution. The random step is followed by a local minimization in torsion angle space. The positions, sizes and preferences for high-probability zones on φ-ψ maps and χ-angle maps were calculated for different residue types from the representative set of 191 and 161 protein 3D-structures, respectively. A fast and precise method to evaluate the electrostatic energy of a protein in solution is developed and combined with the BPMC procedure. The method is based on the modified spherical image charge approximation, efficiently projected onto a molecule of arbitrary shape. Comparison with the finite-difference solutions of the Poisson-Boltzmann equation shows high accuracy for our approach. The BPMC procedure is applied successfully to the structure prediction of 12- and 16-residue synthetic peptides and the determination of protein structure from NMR data, with the immunoglobulin binding domain of streptococcal protein G as an example. The BPMC runs display much better convergence properties than the non-biased simulations. The advantage of a true global optimization procedure for NMR structure determination is its ability to cope with local minima originating from data errors and ambiguities in NMR data."
]
} |
1907.09834 | 2963026274 | We extend the Mobile Server Problem, introduced in SPAA'17, to a model where k identical mobile resources, here named servers, answer requests appearing at points in the Euclidean space. In order to reduce communication costs, the positions of the servers can be adapted by a limited distance m_s per round for each server. The costs are measured similar to the classical Page Migration Problem, i.e., answering a request induces costs proportional to the distance to the nearest server, and moving a server induces costs proportional to the distance multiplied with a weight D. We show that, in our model, no online algorithm can have a constant competitive ratio, i.e., one which is independent of the input length n, even if an augmented moving distance of (1+ )m_s is allowed for the online algorithm. Therefore we investigate a restriction of the power of the adversary dictating the sequence of requests: We demand locality of requests, i.e., that consecutive requests come from points in the Euclidean space with distance bounded by some constant m_c. We show constant lower bounds on the competitiveness in this setting (independent of n, but dependent on k, m_s and m_c). On the positive side, we present a deterministic online algorithm with bounded competitiveness when augmented moving distance and locality of requests is assumed. Our algorithm simulates any given algorithm for the classical k-Page Migration problem as guidance for its servers and extends it by a greedy move of one server in every round. The resulting competitive ratio is polynomial in the number of servers k, the ratio between m_c and m_s, the inverse of the augmentation factor 1 and the competitive ratio of the simulated k-Page Migration algorithm. | In the classical @math -Server Problem as introduced by @cite_8 , @math identical servers are located in a metric space and requests are answered by moving at least one of the servers to the point of the request. The associated costs are equal to the total distance moved. showed that no online algorithm could be better than @math -competitive on every metric with at least @math points. They stated as the @math -Server Conjecture that there is a @math -competitive online algorithm for every metric space. Further, the conjecture is shown to hold for @math and @math where @math is the number of points in the metric space. | {
"cite_N": [
"@cite_8"
],
"mid": [
"2002247760"
],
"abstract": [
"The k-server problem is one of the most fundamental online problems. The problem is to schedule k mobile servers to visit a sequence of points in a metric space with minimum total mileage. The k-server conjecture of Manasse, McGeogh, and Sleator states that there exists a k-competitive online algorithm. The conjecture has been open for over 15 years. The top candidate online algorithm for settling this conjecture is the work function algorithm (WFA) which was shown to have competitive ratio at most 2k - 1. In this paper, we lend support to the conjecture that WFA is in fact k-competitive by proving that it achieves this ratio in several special metric spaces: the line, the star, and all metric spaces with k + 2 points."
]
} |
1907.09834 | 2963026274 | We extend the Mobile Server Problem, introduced in SPAA'17, to a model where k identical mobile resources, here named servers, answer requests appearing at points in the Euclidean space. In order to reduce communication costs, the positions of the servers can be adapted by a limited distance m_s per round for each server. The costs are measured similar to the classical Page Migration Problem, i.e., answering a request induces costs proportional to the distance to the nearest server, and moving a server induces costs proportional to the distance multiplied with a weight D. We show that, in our model, no online algorithm can have a constant competitive ratio, i.e., one which is independent of the input length n, even if an augmented moving distance of (1+ )m_s is allowed for the online algorithm. Therefore we investigate a restriction of the power of the adversary dictating the sequence of requests: We demand locality of requests, i.e., that consecutive requests come from points in the Euclidean space with distance bounded by some constant m_c. We show constant lower bounds on the competitiveness in this setting (independent of n, but dependent on k, m_s and m_c). On the positive side, we present a deterministic online algorithm with bounded competitiveness when augmented moving distance and locality of requests is assumed. Our algorithm simulates any given algorithm for the classical k-Page Migration problem as guidance for its servers and extends it by a greedy move of one server in every round. The resulting competitive ratio is polynomial in the number of servers k, the ratio between m_c and m_s, the inverse of the augmentation factor 1 and the competitive ratio of the simulated k-Page Migration algorithm. | Since its introduction, many algorithms have been designed for special cases of the problem. Most notable is the Double-Coverage Algorithm @cite_10 , which is @math -competitive on trees. For general metrics, the best known result is the Work-Function Algorithm, which is shown to be @math -competitive @cite_3 . Although this algorithm seems generally inefficient in case of runtime and memory, there have been studies showing that an efficient implementation of this algorithm is indeed possible @cite_9 @cite_15 . It was also shown that the algorithm has an optimal competitive ratio of @math on line and star metrics, as well as metrics with @math points @cite_11 . | {
"cite_N": [
"@cite_9",
"@cite_3",
"@cite_15",
"@cite_10",
"@cite_11"
],
"mid": [
"2017441966",
"2077182069",
"2069255833",
"2272870368",
"2963064374"
],
"abstract": [
"This paper provides a systematic study of several proposed measures for online algorithms in the context of a specific problem, namely, the two server problem on three colinear points. Even though the problem is simple, it encapsulates a core challenge in online algorithms which is to balance greediness and adaptability. We examine Competitive Analysis, the Max Max Ratio, the Random Order Ratio, Bijective Analysis and Relative Worst Order Analysis, and determine how these measures compare the Greedy Algorithm, Double Coverage, and Lazy Double Coverage, commonly studied algorithms in the context of server problems. We find that by the Max Max Ratio and Bijective Analysis, Greedy is the best of the three algorithms. Under the other measures, Double Coverage and Lazy Double Coverage are better, though Relative Worst Order Analysis indicates that Greedy is sometimes better. Only Bijective Analysis and Relative Worst Order Analysis indicate that Lazy Double Coverage is better than Double Coverage. Our results also provide the first proof of optimality of an algorithm under Relative Worst Order Analysis.",
"We present the first poly-logarithmic competitive online algorithm for minimum metric bipartite matching. Via induction and a careful use of potential functions, we show that a simple randomized greedy algorithm is competitive on a hierarchically separated tree. Application of recent results on randomized embedding of metrics into trees yield the poly-logarithmic result for general metrics.",
"We prove that the work function algorithm for the k -server problem has a competitive ratio at most 2 k −1. [1988] conjectured that the competitive ratio for the k -server problem is exactly k (it is trivially at least k ); previously the best-known upper bound was exponential in k . Our proof involves three crucial ingredients: A quasiconvexity property of work functions, a duality lemma that uses quasiconvexity to characterize the configuration that achieve maximum increase of the work function, and a potential function that exploits the duality lemma.",
"We consider the problem of online scheduling of jobs on unrelated machines with dynamic speed scaling to minimize the sum of energy and weighted flow time. We give an algorithm with an almost optimal competitive ratio for arbitrary power functions. (No earlier results handled arbitrary power functions for minimizing flow time plus energy with unrelated machines.) For power functions of the form f(s) = sα for some constant α > 1, we get a competitive ratio of O(α log α), improving upon a previous competitive ratio of O(α2) by [3], along with a matching lower bound of Ω(α log α). Further, in the resource augmentation model, with a 1 + e speed up, we give a 2(1 e + 1) competitive algorithm, with essentially the same techniques, improving the bound of 1 + O(1 e2) by [15] and matching the bound of [3] for the special case of fixed speed unrelated machines. Unlike the previous results most of which used an amortized local competitiveness argument or dual fitting methods, we use a primal-dual method, which is useful not only to analyze the algorithms but also to design the algorithm itself.",
"We exhibit a poly(log k)-competitive randomized algorithm for the k-server problem on any metric space. The best previous result independent of the geometry of the underlying metric space is the 2k–1 competitive ratio established for the deterministic work function algorithm by Koutsoupias and Papadimitriou (1995). Even for the special case when the underlying metric space is the real line, the best known competitive ratio was k. Since deterministic algorithms can do no better than k on any metric space with at least k+1 points, this establishes that for every metric space on which the problem is non-trivial, randomized algorithms give an exponential improvement over deterministic algorithms. Our algorithm maintains an approximation of the underlying metric space by a distribution over HSTs. The granularity and accuracy of the approximation is adjusted dynamically according to the aggregate behavior of the HST algorithms. In short: We try to obtain more accurate approximations at the locations and scales where the gactionh is happening. Thus a crucial component of our approach is the O((log k)^2)-competitive randomized algorithm for HSTs obtained in our previous work with Bubeck, Cohen, Lee, and Ma.dry, and its \"multiscale information theory\" perspective."
]
} |
1907.09834 | 2963026274 | We extend the Mobile Server Problem, introduced in SPAA'17, to a model where k identical mobile resources, here named servers, answer requests appearing at points in the Euclidean space. In order to reduce communication costs, the positions of the servers can be adapted by a limited distance m_s per round for each server. The costs are measured similar to the classical Page Migration Problem, i.e., answering a request induces costs proportional to the distance to the nearest server, and moving a server induces costs proportional to the distance multiplied with a weight D. We show that, in our model, no online algorithm can have a constant competitive ratio, i.e., one which is independent of the input length n, even if an augmented moving distance of (1+ )m_s is allowed for the online algorithm. Therefore we investigate a restriction of the power of the adversary dictating the sequence of requests: We demand locality of requests, i.e., that consecutive requests come from points in the Euclidean space with distance bounded by some constant m_c. We show constant lower bounds on the competitiveness in this setting (independent of n, but dependent on k, m_s and m_c). On the positive side, we present a deterministic online algorithm with bounded competitiveness when augmented moving distance and locality of requests is assumed. Our algorithm simulates any given algorithm for the classical k-Page Migration problem as guidance for its servers and extends it by a greedy move of one server in every round. The resulting competitive ratio is polynomial in the number of servers k, the ratio between m_c and m_s, the inverse of the augmentation factor 1 and the competitive ratio of the simulated k-Page Migration algorithm. | The study of randomized online algorithms was initiated by @cite_4 who gave a @math -competitive algorithm for the complete graph. It is speculated that this factor can be obtained for all metrics, however the question is still open. For general metrics, the first algorithm with polylogarithmic competitive ratio was an @math -competitive algorithm by @cite_14 . This was recently improved by @cite_7 who gave an @math -competitive algorithm for HSTs which can be turned into an @math -competitive one for general metrics by a dynamic embedding of general metrics into HSTs @cite_13 . | {
"cite_N": [
"@cite_13",
"@cite_14",
"@cite_4",
"@cite_7"
],
"mid": [
"2963064374",
"2113184918",
"2077182069",
"2114493937"
],
"abstract": [
"We exhibit a poly(log k)-competitive randomized algorithm for the k-server problem on any metric space. The best previous result independent of the geometry of the underlying metric space is the 2k–1 competitive ratio established for the deterministic work function algorithm by Koutsoupias and Papadimitriou (1995). Even for the special case when the underlying metric space is the real line, the best known competitive ratio was k. Since deterministic algorithms can do no better than k on any metric space with at least k+1 points, this establishes that for every metric space on which the problem is non-trivial, randomized algorithms give an exponential improvement over deterministic algorithms. Our algorithm maintains an approximation of the underlying metric space by a distribution over HSTs. The granularity and accuracy of the approximation is adjusted dynamically according to the aggregate behavior of the HST algorithms. In short: We try to obtain more accurate approximations at the locations and scales where the gactionh is happening. Thus a crucial component of our approach is the O((log k)^2)-competitive randomized algorithm for HSTs obtained in our previous work with Bubeck, Cohen, Lee, and Ma.dry, and its \"multiscale information theory\" perspective.",
"We give the first polylogarithmic-competitive randomized online algorithm for the k-server problem on an arbitrary finite metric space. In particular, our algorithm achieves a competitive ratio of O(log3 n log2 k) for any metric space on n points. Our algorithm improves upon the deterministic (2k-1)-competitive algorithm of Koutsoupias and Papadimitriou [Koutsoupias and Papadimitriou 1995] for a wide range of n.",
"We present the first poly-logarithmic competitive online algorithm for minimum metric bipartite matching. Via induction and a careful use of potential functions, we show that a simple randomized greedy algorithm is competitive on a hierarchically separated tree. Application of recent results on randomized embedding of metrics into trees yield the poly-logarithmic result for general metrics.",
"This paper provides a novel technique for the analysis of randomized algorithms for optimization problems on metric spaces, by relating the randomized performance ratio for any, metric space to the randomized performance ratio for a set of \"simple\" metric spaces. We define a notion of a set of metric spaces that probabilistically-approximates another metric space. We prove that any metric space can be probabilistically-approximated by hierarchically well-separated trees (HST) with a polylogarithmic distortion. These metric spaces are \"simple\" as being: (1) tree metrics; (2) natural for applying a divide-and-conquer algorithmic approach. The technique presented is of particular interest in the context of on-line computation. A large number of on-line algorithmic problems, including metrical task systems, server problems, distributed paging, and dynamic storage rearrangement are defined in terms of some metric space. Typically for these problems, there are linear lower bounds on the competitive ratio of deterministic algorithms. Although randomization against an oblivious adversary has the potential of overcoming these high ratios, very little progress has been made in the analysis. We demonstrate the use of our technique by obtaining substantially improved results for two different on-line problems."
]
} |
1907.09834 | 2963026274 | We extend the Mobile Server Problem, introduced in SPAA'17, to a model where k identical mobile resources, here named servers, answer requests appearing at points in the Euclidean space. In order to reduce communication costs, the positions of the servers can be adapted by a limited distance m_s per round for each server. The costs are measured similar to the classical Page Migration Problem, i.e., answering a request induces costs proportional to the distance to the nearest server, and moving a server induces costs proportional to the distance multiplied with a weight D. We show that, in our model, no online algorithm can have a constant competitive ratio, i.e., one which is independent of the input length n, even if an augmented moving distance of (1+ )m_s is allowed for the online algorithm. Therefore we investigate a restriction of the power of the adversary dictating the sequence of requests: We demand locality of requests, i.e., that consecutive requests come from points in the Euclidean space with distance bounded by some constant m_c. We show constant lower bounds on the competitiveness in this setting (independent of n, but dependent on k, m_s and m_c). On the positive side, we present a deterministic online algorithm with bounded competitiveness when augmented moving distance and locality of requests is assumed. Our algorithm simulates any given algorithm for the classical k-Page Migration problem as guidance for its servers and extends it by a greedy move of one server in every round. The resulting competitive ratio is polynomial in the number of servers k, the ratio between m_c and m_s, the inverse of the augmentation factor 1 and the competitive ratio of the simulated k-Page Migration algorithm. | Regarding the Page Migration Problem @cite_2 (also known as File Migration Problem), most results focus on online algorithms which handle only a single page. Contrary to the @math -Server Problem, the design of such algorithms is not trivial for the Page Migration Problem. To the best of our knowledge, the current best results are a @math -competitive deterministic algorithm by @cite_0 and a collection of randomized algorithms with competitive ratio of at most 3 by Jeffery Westbrook @cite_5 . The most relevant results for our problem are two constructions by @cite_1 who give both a deterministic and a randomized algorithm which transform a given algorithm for the @math -Server problem into a deterministic randomized algorithm for the @math -Page Migration Problem. If the given @math -Server algorithm is @math -competitive, the deterministic algorithm is @math -competitive, the randomized algorithm is @math -competitive. Conversely, we use the resulting algorithms as a black box in our constructions. | {
"cite_N": [
"@cite_0",
"@cite_5",
"@cite_1",
"@cite_2"
],
"mid": [
"2092340542",
"2049943172",
"2508179470",
"1486124781"
],
"abstract": [
"This paper is concerned with the page migration (or file migration) problem (Black and Sleator, Technical Report CMU-CS-89-201, Department of Computer Science, Carnegie-Mellon University, 1989) as part of a large class of on-line problems. The page migration problem deals with the management of pages residing in a network of processors. In the classical problem there is only one copy of each page which is accessed by different processors over time. The page is allowed to be migrated between processors. However a migration incurs higher communication cost than an access (proportionally to the page size). The problem is that of deciding when and where to migrate the page in order to lower access costs. A more general setting is the k-page migration problem where we wish to maintain k copies of the page. The page migration problems are concerned with a dilemma common to many on-line problems: determining when it is beneficial to make configuration changes. We deal with the relaxed task systems model which captures a large class of problems of this type, that can be described as the generalization of some original task system problem (, J. ACM 39(4) (1992) 745-763). Given a c-competitive algorithm for a task system we show how to obtain a deterministic O(c2) and randomized O(c) competitive algorithms for the corresponding relaxed task system. The result implies deterministic algorithms for k-page migration by using k-server (, J. Algorithms 11(2) (1990) 208-230) algorithms, and for network leasing by using generalized Steiner tree algorithms (, Proc 7th Ann. ACM-SIAM Symp. on Discrete Algorithms, January 1996, pp. 68-74), as well as providing solutions for natural generalizations of other problems (e.g. storage rearrangement (, Proc. 36th Ann. IEEE Symp. on Foundations of Computer Science, October 1995, pp. 392-403)). We further study some special cases of the k-page migration problem and get optimal deterministic algorithms. For the classical page migration problem we present a deterministic algorithm that achieves a competitive ratio of 4:086, improving upon the previously best competitive ratio of 7 (, Proc. 25th ACM Symp. on Theory of Computing, May 1993, pp. 164-173). (The current lower bound on the problem is 3:148 (, J. Algorithms 24(1) (1997) 124-157). Copyright 2001 Elsevier Science B.V.",
"The page migration problem is to manage a globally addressed shared memory in a multiprocessor system. Each physical page of memory is located at a given processor, and memory references to that page by other processors incur a cost proportional to the network distance. At times the page may migrate between processors at cost proportional to the distance times @math , a page size factor. The problem is to schedule movements on-line so that the total cost of memory references is within a constant factor @math of the best off-line schedule. An algorithm that does so is called c-competitive. Black and Sleator gave 3-competitive deterministic on-line algorithms for uniform networks (complete graphs with unit edge lengths) and for trees with arbitrary edge lengths. No good deterministic algorithm is known for general networks with arbitrary edge lengths. Randomized algorithms are presented for the migration problem that are both simple and better than 3-competitive against an oblivious adversary. An algorithm for uniform graphs is given. It is approximately 2.28-competitive as @math grows large. A second, more powerful algorithm that works on graphs with arbitrary edge distances is also given. This algorithm is approximately 2.62-competitive (or, 1 plus the golden ratio) for large @math . Both these algorithms use random bits only during an initialization phase, and from then on run deterministically. The competitiveness of a very simple coin-flipping algorithm is also examined.",
"In this paper, we construct a deterministic 4-competitive algorithm for the online file migration problem, beating the currently best 20-year old, 4.086-competitive MTLM algorithm by (SODA 1997). Like MTLM, our algorithm also operates in phases, but it adapts their lengths dynamically depending on the geometry of requests seen so far. The improvement was obtained by carefully analyzing a linear model (factor-revealing LP) of a single phase of the algorithm. We also show that if an online algorithm operates in phases of fixed length and the adversary is able to modify the graph between phases, no algorithm can beat the competitive ratio of 4.086.",
"In this paper we consider problems that arise in a shared memory multiprocessor in which memory is physically distributed among a number of memories local to each processor or cluster of processors. The issue we address is that of deciding which local memories should contain copies of pages of data. In the migration problem we operate under the constraint that a page must be kept in exactly one local memory. In the replication problem we allow a page to be kept in any subset of the local memories, but do not allow a local memory to drop a page once it has it. For interconnection topologies that are complete graphs, or trees we have obtained efficient on-line algorithms for these problems. Our migration algorithms also extend to interconnections that are products of these topologies (e.g. a hypercube is a product of simple trees). An on-line algorithm decides how to process each request (which is a read or write request from a processor to a page) without knowing future requests. Our algorithms are also said to be competitive because their performance is within a small constant factor of that of any other algorithm, including algorithms that make use of knowledge of future requests. This research was supported in part by the National Science Foundation under grant CCR8658139. This research was sponsored by the Defense Advanced Research Projects Agency (DOD), monitored by the Space and Naval Warfare Systems Command under Contract N00039-87-C-0251. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Defense Advanced Research Projects Agency or the US Government."
]
} |
1901.09513 | 2913397231 | Underwater robots are subject to position drift due to the effect of ocean currents and the lack of accurate localisation while submerged. We are interested in exploiting such position drift to estimate the ocean current in the surrounding area, thereby assisting navigation and planning. We present a Gaussian process (GP)-based expectation-maximisation (EM) algorithm that estimates the underlying ocean current using sparse GPS data obtained on the surface and dead-reckoned position estimates. We first develop a specialised GP regression scheme that exploits the incompressibility of ocean currents to counteract the underdetermined nature of the problem. We then use the proposed regression scheme in an EM algorithm that estimates the best-fitting ocean current in between each GPS fix. The proposed algorithm is validated in simulation and on a real dataset, and is shown to be capable of reconstructing the underlying ocean current field. We expect to use this algorithm to close the loop between planning and estimation for underwater navigation in unknown ocean currents. | To estimate ocean currents online, one approach is to consider current as a low-frequency disturbance and then apply an extended Kalman filter (EKF) @cite_7 or nonlinear observer @cite_30 in conjunction with acoustic sensors. However, modelling current as a temporal phenomenon clearly overlooks its spatial structure, and acoustic sensors typically require a stationary reference (e.g., the seabed) @cite_13 . | {
"cite_N": [
"@cite_30",
"@cite_13",
"@cite_7"
],
"mid": [
"2438306175",
"2801476520",
"2057235886"
],
"abstract": [
"As applications for autonomous ocean vehicles expand into more dynamic and constrained environments, such as shallow, coastal areas, the benefits of using more precise dynamic model for control and estimation become more compelling. This paper presents a nonlinear observer for current estimation based on AUV dynamic model. Here, AUV dynamic model in currents is taken into consideration. Motivated by the design method of high-gain observer, we take the current disturbances as the uncertainties of the vehicle dynamic system and design the observer gain matrix with the goal of making the observer robust to the effect of current disturbances. The nonlinear observer estimates vehicle's relative velocity firstly; current velocity is further calculated in an indirect way. The proposed current estimation method is validated by numerical simulation.",
"In this paper, a received signal strength (RSS) based localization technique is investigated for underwater optical wireless sensor networks (UOWSNs) where optical noise sources (e.g., sunlight, background, thermal, and dark current) and channel impairments of seawater (e.g., absorption, scattering, and turbulence) pose significant challenges. Hence, we propose a localization technique that works on the noisy ranging measurements embedded in a higher dimensional space and localize the sensor network in a low dimensional space. Once the neighborhood information is measured, a weighted network graph is constructed, which contains the one-hop neighbor distance estimations. A novel approach is developed to complete the missing distances in the kernel matrix. The output of the proposed technique is fused with Helmert transformation to refine the final location estimation with the help of anchors. The simulation results show that the root means square positioning error (RMSPE) of the proposed technique is more robust and accurate compared to baseline and manifold regularization.",
"We extend existing oceanographic sampling methodologies to sample an advecting feature of interest using autonomous robotic platforms. GPS-tracked Lagrangian drifters are used to tag and track a water patch of interest with position updates provided periodically to an autonomous underwater vehicle (AUV) for surveys around the drifter as it moves with ocean currents. Autonomous sampling methods currently rely on geographic waypoint track-line surveys that are suitable for static or slowly changing features. When studying dynamic, rapidly evolving oceanographic features, such methods at best introduce error through insufficient spatial and temporal resolution, and at worst, completely miss the spatial and temporal domain of interest. We demonstrate two approaches for tracking and sampling of advecting oceanographic features. The first relies on extending static-plan AUV surveys (the current state-of-the-art) to sample advecting features. The second approach involves planning of surveys in the drifter or patch frame of reference. We derive a quantitative envelope on patch speeds that can be tracked autonomously by AUVs and drifters and show results from a multi-day off-shore field trial. The results from the trial demonstrate the applicability of our approach to long-term tracking and sampling of advecting features. Additionally, we analyze the data from the trial to identify the sources of error that affect the quality of the surveys carried out. Our work presents the first set of experiments to autonomously observe advecting oceanographic features in the open ocean."
]
} |
1901.09513 | 2913397231 | Underwater robots are subject to position drift due to the effect of ocean currents and the lack of accurate localisation while submerged. We are interested in exploiting such position drift to estimate the ocean current in the surrounding area, thereby assisting navigation and planning. We present a Gaussian process (GP)-based expectation-maximisation (EM) algorithm that estimates the underlying ocean current using sparse GPS data obtained on the surface and dead-reckoned position estimates. We first develop a specialised GP regression scheme that exploits the incompressibility of ocean currents to counteract the underdetermined nature of the problem. We then use the proposed regression scheme in an EM algorithm that estimates the best-fitting ocean current in between each GPS fix. The proposed algorithm is validated in simulation and on a real dataset, and is shown to be capable of reconstructing the underlying ocean current field. We expect to use this algorithm to close the loop between planning and estimation for underwater navigation in unknown ocean currents. | An approach that does consider the spatial nature of the problem is presented in @cite_0 . The authors examine the feasibility of ocean current estimation through simply calculating the average current velocity by dividing the position drift by time. Unsurprisingly, the estimate is increasingly unreliable as distance between diving and surfacing locations grows, and no predictive capability is provided. | {
"cite_N": [
"@cite_0"
],
"mid": [
"2438306175"
],
"abstract": [
"As applications for autonomous ocean vehicles expand into more dynamic and constrained environments, such as shallow, coastal areas, the benefits of using more precise dynamic model for control and estimation become more compelling. This paper presents a nonlinear observer for current estimation based on AUV dynamic model. Here, AUV dynamic model in currents is taken into consideration. Motivated by the design method of high-gain observer, we take the current disturbances as the uncertainties of the vehicle dynamic system and design the observer gain matrix with the goal of making the observer robust to the effect of current disturbances. The nonlinear observer estimates vehicle's relative velocity firstly; current velocity is further calculated in an indirect way. The proposed current estimation method is validated by numerical simulation."
]
} |
1901.09482 | 2913508739 | What is the current state-of-the-art for image restoration and enhancement applied to degraded images acquired under less than ideal circumstances? Can the application of such algorithms as a pre-processing step to improve image interpretability for manual analysis or automatic visual recognition to classify scene content? While there have been important advances in the area of computational photography to restore or enhance the visual quality of an image, the capabilities of such techniques have not always translated in a useful way to visual recognition tasks. Consequently, there is a pressing need for the development of algorithms that are designed for the joint problem of improving visual appearance and recognition, which will be an enabling factor for the deployment of visual recognition tools in many real-world scenarios. To address this, we introduce the UG^2 dataset as a large-scale benchmark composed of video imagery captured under challenging conditions, and two enhancement tasks designed to test algorithmic impact on visual quality and automatic object recognition. Furthermore, we propose a set of metrics to evaluate the joint improvement of such tasks as well as individual algorithmic advances, including a novel psychophysics-based evaluation regime for human assessment and a realistic set of quantitative measures for object recognition performance. We introduce six new algorithms for image restoration or enhancement, which were created as part of the IARPA sponsored UG^2 Challenge workshop held at CVPR 2018. Under the proposed evaluation regime, we present an in-depth analysis of these algorithms and a host of deep learning-based and classic baseline approaches. From the observed results, it is evident that we are in the early days of building a bridge between computational photography and visual recognition, leaving many opportunities for innovation in this area. | The areas of image restoration and enhancement have a long history in computational photography, with associated benchmark datasets that are mainly used for the qualitative evaluation of image appearance. These include very small test image sets such as Set5 @cite_130 and Set14 @cite_44 , the set of blurred images introduced by Levin al @cite_31 , and the DIVerse 2K resolution image dataset (DIV2K) @cite_55 designed for super-resolution benchmarking. Datasets containing more diverse scene content have been proposed including Urban100 @cite_105 for enhancement comparisons and LIVE1 @cite_125 for image quality assessment. While not originally designed for computational photography, the Berkeley Segmentation Dataset has been used by itself @cite_105 and in combination with LIVE1 @cite_123 for enhancement work. The popularity of deep learning methods has increased demand for training and testing data, which Su al provide as video content for deblurring work @cite_99 . Importantly, none of these datasets were designed to combine image restoration and enhancement with recognition for a unified benchmark. | {
"cite_N": [
"@cite_99",
"@cite_55",
"@cite_125",
"@cite_130",
"@cite_44",
"@cite_31",
"@cite_123",
"@cite_105"
],
"mid": [
"2523714292",
"2963470893",
"2951706743",
"2055686029",
"2134557905",
"2754419751",
"2963068450",
"2148848374"
],
"abstract": [
"Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors? The behavior of optimization-based super-resolution methods is principally driven by the choice of the objective function. Recent work has largely focused on minimizing the mean squared reconstruction error. The resulting estimates have high peak signal-to-noise ratios, but they are often lacking high-frequency details and are perceptually unsatisfying in the sense that they fail to match the fidelity expected at the higher resolution. In this paper, we present SRGAN, a generative adversarial network (GAN) for image super-resolution (SR). To our knowledge, it is the first framework capable of inferring photo-realistic natural images for 4x upscaling factors. To achieve this, we propose a perceptual loss function which consists of an adversarial loss and a content loss. The adversarial loss pushes our solution to the natural image manifold using a discriminator network that is trained to differentiate between the super-resolved images and original photo-realistic images. In addition, we use a content loss motivated by perceptual similarity instead of similarity in pixel space. Our deep residual network is able to recover photo-realistic textures from heavily downsampled images on public benchmarks. An extensive mean-opinion-score (MOS) test shows hugely significant gains in perceptual quality using SRGAN. The MOS scores obtained with SRGAN are closer to those of the original high-resolution images than to those obtained with any state-of-the-art method.",
"Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors? The behavior of optimization-based super-resolution methods is principally driven by the choice of the objective function. Recent work has largely focused on minimizing the mean squared reconstruction error. The resulting estimates have high peak signal-to-noise ratios, but they are often lacking high-frequency details and are perceptually unsatisfying in the sense that they fail to match the fidelity expected at the higher resolution. In this paper, we present SRGAN, a generative adversarial network (GAN) for image super-resolution (SR). To our knowledge, it is the first framework capable of inferring photo-realistic natural images for 4x upscaling factors. To achieve this, we propose a perceptual loss function which consists of an adversarial loss and a content loss. The adversarial loss pushes our solution to the natural image manifold using a discriminator network that is trained to differentiate between the super-resolved images and original photo-realistic images. In addition, we use a content loss motivated by perceptual similarity instead of similarity in pixel space. Our deep residual network is able to recover photo-realistic textures from heavily downsampled images on public benchmarks. An extensive mean-opinion-score (MOS) test shows hugely significant gains in perceptual quality using SRGAN. The MOS scores obtained with SRGAN are closer to those of the original high-resolution images than to those obtained with any state-of-the-art method.",
"With deep learning becoming the dominant approach in computer vision, the use of representations extracted from Convolutional Neural Nets (CNNs) is quickly gaining ground on Fisher Vectors (FVs) as favoured state-of-the-art global image descriptors for image instance retrieval. While the good performance of CNNs for image classification are unambiguously recognised, which of the two has the upper hand in the image retrieval context is not entirely clear yet. In this work, we propose a comprehensive study that systematically evaluates FVs and CNNs for image retrieval. The first part compares the performances of FVs and CNNs on multiple publicly available data sets. We investigate a number of details specific to each method. For FVs, we compare sparse descriptors based on interest point detectors with dense single-scale and multi-scale variants. For CNNs, we focus on understanding the impact of depth, architecture and training data on retrieval results. Our study shows that no descriptor is systematically better than the other and that performance gains can usually be obtained by using both types together. The second part of the study focuses on the impact of geometrical transformations such as rotations and scale changes. FVs based on interest point detectors are intrinsically resilient to such transformations while CNNs do not have a built-in mechanism to ensure such invariance. We show that performance of CNNs can quickly degrade in presence of rotations while they are far less affected by changes in scale. We then propose a number of ways to incorporate the required invariances in the CNN pipeline. Overall, our work is intended as a reference guide offering practically useful and simply implementable guidelines to anyone looking for state-of-the-art global descriptors best suited to their specific image instance retrieval problem.",
"We propose a probabilistic formulation of joint silhouette extraction and 3D reconstruction given a series of calibrated 2D images. Instead of segmenting each image separately in order to construct a 3D surface consistent with the estimated silhouettes, we compute the most probable 3D shape that gives rise to the observed color information. The probabilistic framework, based on Bayesian inference, enables robust 3D reconstruction by optimally taking into account the contribution of all views. We solve the arising maximum a posteriori shape inference in a globally optimal manner by convex relaxation techniques in a spatially continuous representation. For an interactively provided user input in the form of scribbles specifying foreground and background regions, we build corresponding color distributions as multivariate Gaussians and find a volume occupancy that best fits to this data in a variational sense. Compared to classical methods for silhouette-based multiview reconstruction, the proposed approach does not depend on initialization and enjoys significant resilience to violations of the model assumptions due to background clutter, specular reflections, and camera sensor perturbations. In experiments on several real-world data sets, we show that exploiting a silhouette coherency criterion in a multiview setting allows for dramatic improvements of silhouette quality over independent 2D segmentations without any significant increase of computational efforts. This results in more accurate visual hull estimation, needed by a multitude of image-based modeling approaches. We made use of recent advances in parallel computing with a GPU implementation of the proposed method generating reconstructions on volume grids of more than 20 million voxels in up to 4.41 seconds.",
"We assess the applicability of several popular learning methods for the problem of recognizing generic visual categories with invariance to pose, lighting, and surrounding clutter. A large dataset comprising stereo image pairs of 50 uniform-colored toys under 36 azimuths, 9 elevations, and 6 lighting conditions was collected (for a total of 194,400 individual images). The objects were 10 instances of 5 generic categories: four-legged animals, human figures, airplanes, trucks, and cars. Five instances of each category were used for training, and the other five for testing. Low-resolution grayscale images of the objects with various amounts of variability and surrounding clutter were used for training and testing. Nearest neighbor methods, support vector machines, and convolutional networks, operating on raw pixels or on PCA-derived features were tested. Test error rates for unseen object instances placed on uniform backgrounds were around 13 for SVM and 7 for convolutional nets. On a segmentation recognition task with highly cluttered images, SVM proved impractical, while convolutional nets yielded 16 7 error. A real-time version of the system was implemented that can detect and classify objects in natural scenes at around 10 frames per second.",
"The conventional methods for estimating camera poses and scene structures from severely blurry or low resolution images often result in failure. The off-the-shelf deblurring or super-resolution methods may show visually pleasing results. However, applying each technique independently before matching is generally unprofitable because this naive series of procedures ignores the consistency between images. In this paper, we propose a pioneering unified framework that solves four problems simultaneously, namely, dense depth reconstruction, camera pose estimation, super-resolution, and deblurring. By reflecting a physical imaging process, we formulate a cost minimization problem and solve it using an alternating optimization technique. The experimental results on both synthetic and real videos show high-quality depth maps derived from severely degraded images that contrast the failures of naive multi-view stereo methods. Our proposed method also produces outstanding deblurred and super-resolved images unlike the independent application or combination of conventional video deblurring, super-resolution methods.",
"Convolutional neural networks rely on image texture and structure to serve as discriminative features to classify the image content. Image enhancement techniques can be used as preprocessing steps to help improve the overall image quality and in turn improve the overall effectiveness of a CNN. Existing image enhancement methods, however, are designed to improve the perceptual quality of an image for a human observer. In this paper, we are interested in learning CNNs that can emulate image enhancement and restoration, but with the overall goal to improve image classification and not necessarily human perception. To this end, we present a unified CNN architecture that uses a range of enhancement filters that can enhance image-specific details via end-to-end dynamic filter learning. We demonstrate the effectiveness of this strategy on four challenging benchmark datasets for fine-grained, object, scene, and texture classification: CUB-200-2011, PASCAL-VOC2007, MIT-Indoor, and DTD. Experiments using our proposed enhancement show promising results on all the datasets. In addition, our approach is capable of improving the performance of all generic CNN architectures.",
"Most publicly available image quality databases have been created under highly controlled conditions by introducing graded simulated distortions onto high-quality photographs. However, images captured using typical real-world mobile camera devices are usually afflicted by complex mixtures of multiple distortions, which are not necessarily well-modeled by the synthetic distortions found in existing databases. The originators of existing legacy databases usually conducted human psychometric studies to obtain statistically meaningful sets of human opinion scores on images in a stringently controlled visual environment, resulting in small data collections relative to other kinds of image analysis databases. Toward overcoming these limitations, we designed and created a new database that we call the LIVE In the Wild Image Quality Challenge Database, which contains widely diverse authentic image distortions on a large number of images captured using a representative variety of modern mobile devices. We also designed and implemented a new online crowdsourcing system, which we have used to conduct a very large-scale, multi-month image quality assessment (IQA) subjective study. Our database consists of over 350 000 opinion scores on 1162 images evaluated by over 8100 unique human observers. Despite the lack of control over the experimental environments of the numerous study participants, we demonstrate excellent internal consistency of the subjective data set. We also evaluate several top-performing blind IQA algorithms on it and present insights on how the mixtures of distortions challenge both end users as well as automatic perceptual quality prediction models. The new database is available for public use at http: live.ece.utexas.edu research ChallengeDB index.html ."
]
} |
1901.09482 | 2913508739 | What is the current state-of-the-art for image restoration and enhancement applied to degraded images acquired under less than ideal circumstances? Can the application of such algorithms as a pre-processing step to improve image interpretability for manual analysis or automatic visual recognition to classify scene content? While there have been important advances in the area of computational photography to restore or enhance the visual quality of an image, the capabilities of such techniques have not always translated in a useful way to visual recognition tasks. Consequently, there is a pressing need for the development of algorithms that are designed for the joint problem of improving visual appearance and recognition, which will be an enabling factor for the deployment of visual recognition tools in many real-world scenarios. To address this, we introduce the UG^2 dataset as a large-scale benchmark composed of video imagery captured under challenging conditions, and two enhancement tasks designed to test algorithmic impact on visual quality and automatic object recognition. Furthermore, we propose a set of metrics to evaluate the joint improvement of such tasks as well as individual algorithmic advances, including a novel psychophysics-based evaluation regime for human assessment and a realistic set of quantitative measures for object recognition performance. We introduce six new algorithms for image restoration or enhancement, which were created as part of the IARPA sponsored UG^2 Challenge workshop held at CVPR 2018. Under the proposed evaluation regime, we present an in-depth analysis of these algorithms and a host of deep learning-based and classic baseline approaches. From the observed results, it is evident that we are in the early days of building a bridge between computational photography and visual recognition, leaving many opportunities for innovation in this area. | With respect to data collected by aerial vehicles, the VIRAT Video Dataset @cite_80 contains realistic, natural and challenging (in terms of its resolution, background clutter, diversity in scenes)" imagery for event recognition, while the VisDrone2018 Dataset @cite_4 is designed for object detection and tracking. Other datasets including aerial imagery are the UCF Aerial Action Data Set @cite_6 , UCF-ARG @cite_114 , UAV123 @cite_0 , and the multi-purpose dataset introduced by Yao al @cite_36 . As with the computational photography datasets, none of these sets have protocols for image restoration and enhancement coupled with object recognition. | {
"cite_N": [
"@cite_4",
"@cite_36",
"@cite_6",
"@cite_0",
"@cite_80",
"@cite_114"
],
"mid": [
"2798799804",
"2142996775",
"2884821995",
"2314029052",
"2612465689",
"2095537868"
],
"abstract": [
"In this paper we present a large-scale visual object detection and tracking benchmark, named VisDrone2018, aiming at advancing visual understanding tasks on the drone platform. The images and video sequences in the benchmark were captured over various urban suburban areas of 14 different cities across China from north to south. Specifically, VisDrone2018 consists of 263 video clips and 10,209 images (no overlap with video clips) with rich annotations, including object bounding boxes, object categories, occlusion, truncation ratios, etc. With intensive amount of effort, our benchmark has more than 2.5 million annotated instances in 179,264 images video frames. Being the largest such dataset ever published, the benchmark enables extensive evaluation and investigation of visual analysis algorithms on the drone platform. In particular, we design four popular tasks with the benchmark, including object detection in images, object detection in videos, single object tracking, and multi-object tracking. All these tasks are extremely challenging in the proposed dataset due to factors such as occlusion, large scale and pose variation, and fast motion. We hope the benchmark largely boost the research and development in visual analysis on drone platforms.",
"We introduce a new large-scale video dataset designed to assess the performance of diverse visual event recognition algorithms with a focus on continuous visual event recognition (CVER) in outdoor areas with wide coverage. Previous datasets for action recognition are unrealistic for real-world surveillance because they consist of short clips showing one action by one individual [15, 8]. Datasets have been developed for movies [11] and sports [12], but, these actions and scene conditions do not apply effectively to surveillance videos. Our dataset consists of many outdoor scenes with actions occurring naturally by non-actors in continuously captured videos of the real world. The dataset includes large numbers of instances for 23 event types distributed throughout 29 hours of video. This data is accompanied by detailed annotations which include both moving object tracks and event examples, which will provide solid basis for large-scale evaluation. Additionally, we propose different types of evaluation modes for visual recognition tasks and evaluation metrics along with our preliminary experimental results. We believe that this dataset will stimulate diverse aspects of computer vision research and help us to advance the CVER tasks in the years ahead.",
"Abstract Aerial image classification is of great significance in the remote sensing community, and many researches have been conducted over the past few years. Among these studies, most of them focus on categorizing an image into one semantic label, while in the real world, an aerial image is often associated with multiple labels, e.g., multiple object-level labels in our case. Besides, a comprehensive picture of present objects in a given high-resolution aerial image can provide a more in-depth understanding of the studied region. For these reasons, aerial image multi-label classification has been attracting increasing attention. However, one common limitation shared by existing methods in the community is that the co-occurrence relationship of various classes, so-called class dependency, is underexplored and leads to an inconsiderate decision. In this paper, we propose a novel end-to-end network, namely class-wise attention-based convolutional and bidirectional LSTM network (CA-Conv-BiLSTM), for this task. The proposed network consists of three indispensable components: (1) a feature extraction module, (2) a class attention learning layer, and (3) a bidirectional LSTM-based sub-network. Particularly, the feature extraction module is designed for extracting fine-grained semantic feature maps, while the class attention learning layer aims at capturing discriminative class-specific features. As the most important part, the bidirectional LSTM-based sub-network models the underlying class dependency in both directions and produce structured multiple object labels. Experimental results on UCM multi-label dataset and DFC15 multi-label dataset validate the effectiveness of our model quantitatively and qualitatively.",
"We are witnessing daily acquisition of large amounts of aerial and satellite imagery. Analysis of such large quantities of data can be helpful for many practical applications. In this letter, we present an automatic content-based analysis of aerial imagery in order to detect and mark arbitrary objects or regions in high-resolution images. For that purpose, we proposed a method for automatic object detection based on a convolutional neural network. A novel two-stage approach for network training is implemented and verified in the tasks of aerial image classification and object detection. First, we tested the proposed training approach using UCMerced data set of aerial images and achieved accuracy of approximately 98.6 . Second, the method for automatic object detection was implemented and verified. For implementation on GPGPU, a required processing time for one aerial image of size 5000 @math 5000 pixels was around 30 s.",
"Vehicle detection in aerial images is a crucial image processing step for many applications like screening of large areas. In recent years, several deep learning based frameworks have been proposed for object detection. However, these detectors were developed for datasets that considerably differ from aerial images. In this paper, we systematically investigate the potential of Fast R-CNN and Faster R-CNN for aerial images, which achieve top performing results on common detection benchmark datasets. Therefore, the applicability of 8 state-of-the-art object proposals methods used to generate a set of candidate regions and of both detectors is examined. Relevant adaptations of the object proposals methods are provided. To overcome shortcomings of the original approach in case of handling small instances, we further propose our own network that clearly outperforms state-of-the-art methods for vehicle detection in aerial images. All experiments are performed on two publicly available datasets to account for differing characteristics such as ground sampling distance, number of objects per image and varying backgrounds.",
"This paper presents a solution to solve the car detection and counting problem in images acquired by means of unmanned aerial vehicles (UAVs). UAV images are characterized by a very high spatial resolution (order of few centimeters), and consequently by an extremely high level of details which calls for appropriate automatic analysis methods. The proposed method starts with a screening step of asphalted zones in order to restrict the areas where to detect cars and thus to reduce false alarms. Then, it performs a feature extraction process based on scalar invariant feature transform thanks to which a set of keypoints is identified in the considered image and opportunely described. Successively, it discriminates between keypoints assigned to cars and all the others, by means of a support vector machine classifier. The last step of our method is focused on the grouping of the keypoints belonging to the same car in order to get a “one keypoint-one car” relationship. Finally, the number of cars present in the scene is given by the number of final keypoints identified. The experimental results obtained on a real UAV scene characterized by a spatial resolution of 2 cm show that the proposed method exhibits a promising car counting accuracy."
]
} |
1901.09482 | 2913508739 | What is the current state-of-the-art for image restoration and enhancement applied to degraded images acquired under less than ideal circumstances? Can the application of such algorithms as a pre-processing step to improve image interpretability for manual analysis or automatic visual recognition to classify scene content? While there have been important advances in the area of computational photography to restore or enhance the visual quality of an image, the capabilities of such techniques have not always translated in a useful way to visual recognition tasks. Consequently, there is a pressing need for the development of algorithms that are designed for the joint problem of improving visual appearance and recognition, which will be an enabling factor for the deployment of visual recognition tools in many real-world scenarios. To address this, we introduce the UG^2 dataset as a large-scale benchmark composed of video imagery captured under challenging conditions, and two enhancement tasks designed to test algorithmic impact on visual quality and automatic object recognition. Furthermore, we propose a set of metrics to evaluate the joint improvement of such tasks as well as individual algorithmic advances, including a novel psychophysics-based evaluation regime for human assessment and a realistic set of quantitative measures for object recognition performance. We introduce six new algorithms for image restoration or enhancement, which were created as part of the IARPA sponsored UG^2 Challenge workshop held at CVPR 2018. Under the proposed evaluation regime, we present an in-depth analysis of these algorithms and a host of deep learning-based and classic baseline approaches. From the observed results, it is evident that we are in the early days of building a bridge between computational photography and visual recognition, leaving many opportunities for innovation in this area. | Intuitively, if an image has been corrupted, then employing restoration techniques should improve performance of recognizing objects in the image. An early attempt at unifying a high-level task like object recognition with a low-level task like deblurring was performed by Zeiler al through deconvolutional networks @cite_23 @cite_14 . Similarly, Haris al @cite_32 proposed an end-to-end super resolution training procedure that incorporated detection loss as a training objective, obtaining superior object detection results compared to traditional super-resolution methods for a variety of conditions (including additional perturbations on the low resolution images such as the addition of Gaussian noise). | {
"cite_N": [
"@cite_14",
"@cite_32",
"@cite_23"
],
"mid": [
"2523714292",
"2963470893",
"2780624730"
],
"abstract": [
"Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors? The behavior of optimization-based super-resolution methods is principally driven by the choice of the objective function. Recent work has largely focused on minimizing the mean squared reconstruction error. The resulting estimates have high peak signal-to-noise ratios, but they are often lacking high-frequency details and are perceptually unsatisfying in the sense that they fail to match the fidelity expected at the higher resolution. In this paper, we present SRGAN, a generative adversarial network (GAN) for image super-resolution (SR). To our knowledge, it is the first framework capable of inferring photo-realistic natural images for 4x upscaling factors. To achieve this, we propose a perceptual loss function which consists of an adversarial loss and a content loss. The adversarial loss pushes our solution to the natural image manifold using a discriminator network that is trained to differentiate between the super-resolved images and original photo-realistic images. In addition, we use a content loss motivated by perceptual similarity instead of similarity in pixel space. Our deep residual network is able to recover photo-realistic textures from heavily downsampled images on public benchmarks. An extensive mean-opinion-score (MOS) test shows hugely significant gains in perceptual quality using SRGAN. The MOS scores obtained with SRGAN are closer to those of the original high-resolution images than to those obtained with any state-of-the-art method.",
"Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors? The behavior of optimization-based super-resolution methods is principally driven by the choice of the objective function. Recent work has largely focused on minimizing the mean squared reconstruction error. The resulting estimates have high peak signal-to-noise ratios, but they are often lacking high-frequency details and are perceptually unsatisfying in the sense that they fail to match the fidelity expected at the higher resolution. In this paper, we present SRGAN, a generative adversarial network (GAN) for image super-resolution (SR). To our knowledge, it is the first framework capable of inferring photo-realistic natural images for 4x upscaling factors. To achieve this, we propose a perceptual loss function which consists of an adversarial loss and a content loss. The adversarial loss pushes our solution to the natural image manifold using a discriminator network that is trained to differentiate between the super-resolved images and original photo-realistic images. In addition, we use a content loss motivated by perceptual similarity instead of similarity in pixel space. Our deep residual network is able to recover photo-realistic textures from heavily downsampled images on public benchmarks. An extensive mean-opinion-score (MOS) test shows hugely significant gains in perceptual quality using SRGAN. The MOS scores obtained with SRGAN are closer to those of the original high-resolution images than to those obtained with any state-of-the-art method.",
"We present an algorithm to directly restore a clear highresolution image from a blurry low-resolution input. This problem is highly ill-posed and the basic assumptions for existing super-resolution methods (requiring clear input) and deblurring methods (requiring high-resolution input) no longer hold. We focus on face and text images and adopt a generative adversarial network (GAN) to learn a category-specific prior to solve this problem. However, the basic GAN formulation does not generate realistic highresolution images. In this work, we introduce novel training losses that help recover fine details. We also present a multi-class GAN that can process multi-class image restoration tasks, i.e., face and text images, using a single generator network. Extensive experiments demonstrate that our method performs favorably against the state-of-the-art methods on both synthetic and real-world images at a lower computational cost."
]
} |
1901.09482 | 2913508739 | What is the current state-of-the-art for image restoration and enhancement applied to degraded images acquired under less than ideal circumstances? Can the application of such algorithms as a pre-processing step to improve image interpretability for manual analysis or automatic visual recognition to classify scene content? While there have been important advances in the area of computational photography to restore or enhance the visual quality of an image, the capabilities of such techniques have not always translated in a useful way to visual recognition tasks. Consequently, there is a pressing need for the development of algorithms that are designed for the joint problem of improving visual appearance and recognition, which will be an enabling factor for the deployment of visual recognition tools in many real-world scenarios. To address this, we introduce the UG^2 dataset as a large-scale benchmark composed of video imagery captured under challenging conditions, and two enhancement tasks designed to test algorithmic impact on visual quality and automatic object recognition. Furthermore, we propose a set of metrics to evaluate the joint improvement of such tasks as well as individual algorithmic advances, including a novel psychophysics-based evaluation regime for human assessment and a realistic set of quantitative measures for object recognition performance. We introduce six new algorithms for image restoration or enhancement, which were created as part of the IARPA sponsored UG^2 Challenge workshop held at CVPR 2018. Under the proposed evaluation regime, we present an in-depth analysis of these algorithms and a host of deep learning-based and classic baseline approaches. From the observed results, it is evident that we are in the early days of building a bridge between computational photography and visual recognition, leaving many opportunities for innovation in this area. | Sajjadi al @cite_87 argue that the use of traditional metrics such as Peak Signal to Noise Ratio (PSNR), Structural Similarity Index (SSIM), or the Information Fidelity Criterion (IFC) might not reflect the performance of some models, and propose the use of object recognition performance as an evaluation metric. They observed that methods that produced images of higher perceptual quality obtained higher classification performance despite obtaining low PSNR scores. In agreement with this, Gondal al @cite_106 observed the correlation of the perceptual quality of an image with its performance when processed by object recognition models. Similarly, Tahboub al @cite_57 evaluate the impact of degradation caused by video compression on pedestrian detection. Other approaches have used visual recognition as a way to evaluate the performance of visual enhancement algorithm for tasks such as text deblurring @cite_88 @cite_8 , image colorization @cite_33 , and single image super resolution @cite_100 . | {
"cite_N": [
"@cite_33",
"@cite_8",
"@cite_87",
"@cite_57",
"@cite_106",
"@cite_88",
"@cite_100"
],
"mid": [
"2064076387",
"2898677683",
"1984043993",
"2132549992",
"2133946027",
"2174358748",
"2134557905"
],
"abstract": [
"In this paper, we analyse two well-known objective image quality metrics, the peak-signal-to-noise ratio (PSNR) as well as the structural similarity index measure (SSIM), and we derive a simple mathematical relationship between them which works for various kinds of image degradations such as Gaussian blur, additive Gaussian white noise, jpeg and jpeg2000 compression. A series of tests realized on images extracted from the Kodak database gives a better understanding of the similarity and difference between the SSIM and the PSNR.",
"Convolutional neural network (CNN) based methods have recently achieved great success for image super-resolution (SR). However, most deep CNN based SR models attempt to improve distortion measures (e.g. PSNR, SSIM, IFC, VIF) while resulting in poor quantified perceptual quality (e.g. human opinion score, no-reference quality measures such as NIQE). Few works have attempted to improve the perceptual quality at the cost of performance reduction in distortion measures. A very recent study has revealed that distortion and perceptual quality are at odds with each other and there is always a trade-off between the two. Often the restoration algorithms that are superior in terms of perceptual quality, are inferior in terms of distortion measures. Our work attempts to analyze the trade-off between distortion and perceptual quality for the problem of single image SR. To this end, we use the well-known SR architecture- enhanced deep super-resolution (EDSR) network and show that it can be adapted to achieve better perceptual quality for a specific range of the distortion measure. While the original network of EDSR was trained to minimize the error defined based on per-pixel accuracy alone, we train our network using a generative adversarial network framework with EDSR as the generator module. Our proposed network, called enhanced perceptual super-resolution network (EPSR), is trained with a combination of mean squared error loss, perceptual loss, and adversarial loss. Our experiments reveal that EPSR achieves the state-of-the-art trade-off between distortion and perceptual quality while the existing methods perform well in either of these measures alone.",
"Currently two evaluation methods of super-resolution (SR) techniques prevail: The objective Peak Signal to Noise Ratio (PSNR) and a qualitative measure based on manual visual inspection. Both of these methods are sub-optimal: The latter does not scale well to large numbers of images, while the former does not necessarily reflect the perceived visual quality. We address these issues in this paper and propose an evaluation method based on image classification. We show that perceptual image quality measures like structural similarity are not suitable for evaluation of SR methods. On the other hand a systematic evaluation using large datasets of thousands of real-world images provides a consistent comparison of SR algorithms that corresponds to perceived visual quality. We verify the success of our approach by presenting an evaluation of three recent super-resolution algorithms on standard image classification datasets.",
"With the increasing demand for video-based applications, the reliable prediction of video quality has increased in importance. Numerous video quality assessment methods and metrics have been proposed over the past years with varying computational complexity and accuracy. In this paper, we introduce a classification scheme for full-reference and reduced-reference media-layer objective video quality assessment methods. Our classification scheme first classifies a method according to whether natural visual characteristics or perceptual (human visual system) characteristics are considered. We further subclassify natural visual characteristics methods into methods based on natural visual statistics or natural visual features. We subclassify perceptual characteristics methods into frequency or pixel-domain methods. According to our classification scheme, we comprehensively review and compare the media-layer objective video quality models for both standard resolution and high definition video. We find that the natural visual statistics based MultiScale-Structural SIMilarity index (MS-SSIM), the natural visual feature based Video Quality Metric (VQM), and the perceptual spatio-temporal frequency-domain based MOtion-based Video Integrity Evaluation (MOVIE) index give the best performance for the LIVE Video Quality Database.",
"Since the human visual system (HVS) is the ultimate assessor of image quality, current research on the design of objective image quality metrics tends to include an important feature of the HVS, namely, visual attention. Different metrics for image quality prediction have been extended with a computational model of visual attention, but the resulting gain in reliability of the metrics so far was variable. To better understand the basic added value of including visual attention in the design of objective metrics, we used measured data of visual attention. To this end, we performed two eye-tracking experiments: one with a free-looking task and one with a quality assessment task. In the first experiment, 20 observers looked freely to 29 unimpaired original images, yielding us so-called natural scene saliency (NSS). In the second experiment, 20 different observers assessed the quality of distorted versions of the original images. The resulting saliency maps showed some differences with the NSS, and therefore, we applied both types of saliency to four different objective metrics predicting the quality of JPEG compressed images. For both types of saliency the performance gain of the metrics improved, but to a larger extent when adding the NSS. As a consequence, we further integrated NSS in several state-of-the-art quality metrics, including three full-reference metrics and two no-reference metrics, and evaluated their prediction performance for a larger set of distortions. By doing so, we evaluated whether and to what extent the addition of NSS is beneficial to objective quality prediction in general terms. In addition, we address some practical issues in the design of an attention-based metric. The eye-tracking data are made available to the research community .",
"Deep networks are increasingly being applied to problems involving image synthesis, e.g., generating images from textual descriptions and reconstructing an input image from a compact representation. Supervised training of image-synthesis networks typically uses a pixel-wise loss (PL) to indicate the mismatch between a generated image and its corresponding target image. We propose instead to use a loss function that is better calibrated to human perceptual judgments of image quality: the multiscale structural-similarity score (MS-SSIM). Because MS-SSIM is differentiable, it is easily incorporated into gradient-descent learning. We compare the consequences of using MS-SSIM versus PL loss on training deterministic and stochastic autoencoders. For three different architectures, we collected human judgments of the quality of image reconstructions. Observers reliably prefer images synthesized by MS-SSIM-optimized models over those synthesized by PL-optimized models, for two distinct PL measures ( @math and @math distances). We also explore the effect of training objective on image encoding and analyze conditions under which perceptually-optimized representations yield better performance on image classification. Finally, we demonstrate the superiority of perceptually-optimized networks for super-resolution imaging. Just as computer vision has advanced through the use of convolutional architectures that mimic the structure of the mammalian visual system, we argue that significant additional advances can be made in modeling images through the use of training objectives that are well aligned to characteristics of human perception.",
"We assess the applicability of several popular learning methods for the problem of recognizing generic visual categories with invariance to pose, lighting, and surrounding clutter. A large dataset comprising stereo image pairs of 50 uniform-colored toys under 36 azimuths, 9 elevations, and 6 lighting conditions was collected (for a total of 194,400 individual images). The objects were 10 instances of 5 generic categories: four-legged animals, human figures, airplanes, trucks, and cars. Five instances of each category were used for training, and the other five for testing. Low-resolution grayscale images of the objects with various amounts of variability and surrounding clutter were used for training and testing. Nearest neighbor methods, support vector machines, and convolutional networks, operating on raw pixels or on PCA-derived features were tested. Test error rates for unseen object instances placed on uniform backgrounds were around 13 for SVM and 7 for convolutional nets. On a segmentation recognition task with highly cluttered images, SVM proved impractical, while convolutional nets yielded 16 7 error. A real-time version of the system was implemented that can detect and classify objects in natural scenes at around 10 frames per second."
]
} |
1901.09482 | 2913508739 | What is the current state-of-the-art for image restoration and enhancement applied to degraded images acquired under less than ideal circumstances? Can the application of such algorithms as a pre-processing step to improve image interpretability for manual analysis or automatic visual recognition to classify scene content? While there have been important advances in the area of computational photography to restore or enhance the visual quality of an image, the capabilities of such techniques have not always translated in a useful way to visual recognition tasks. Consequently, there is a pressing need for the development of algorithms that are designed for the joint problem of improving visual appearance and recognition, which will be an enabling factor for the deployment of visual recognition tools in many real-world scenarios. To address this, we introduce the UG^2 dataset as a large-scale benchmark composed of video imagery captured under challenging conditions, and two enhancement tasks designed to test algorithmic impact on visual quality and automatic object recognition. Furthermore, we propose a set of metrics to evaluate the joint improvement of such tasks as well as individual algorithmic advances, including a novel psychophysics-based evaluation regime for human assessment and a realistic set of quantitative measures for object recognition performance. We introduce six new algorithms for image restoration or enhancement, which were created as part of the IARPA sponsored UG^2 Challenge workshop held at CVPR 2018. Under the proposed evaluation regime, we present an in-depth analysis of these algorithms and a host of deep learning-based and classic baseline approaches. From the observed results, it is evident that we are in the early days of building a bridge between computational photography and visual recognition, leaving many opportunities for innovation in this area. | While the above approaches employ object recognition in addition to visual enhancement, there are approaches designed to overlook the visual appearance of the image and instead make use of enhancement techniques to exclusively improve the object recognition performance. Sharma al @cite_79 make use of dynamic enhancement filters in an end-to-end processing and classification pipeline that incorporates two loss functions (enhancement and classification). The approach focuses on improving the performance of challenging high quality images. In contrast to this, Yim al @cite_13 propose a classification architecture (comprised of a pre-processing module and a neural network model) to handle images degraded by noise. Li al @cite_40 introduced a dehazing method that is concatenated with Faster R-CNN and jointly optimized as a unified pipeline. It outperforms traditional Faster R-CNN and other non-joint approaches. | {
"cite_N": [
"@cite_79",
"@cite_40",
"@cite_13"
],
"mid": [
"2963068450",
"2964019666",
"2253171278"
],
"abstract": [
"Convolutional neural networks rely on image texture and structure to serve as discriminative features to classify the image content. Image enhancement techniques can be used as preprocessing steps to help improve the overall image quality and in turn improve the overall effectiveness of a CNN. Existing image enhancement methods, however, are designed to improve the perceptual quality of an image for a human observer. In this paper, we are interested in learning CNNs that can emulate image enhancement and restoration, but with the overall goal to improve image classification and not necessarily human perception. To this end, we present a unified CNN architecture that uses a range of enhancement filters that can enhance image-specific details via end-to-end dynamic filter learning. We demonstrate the effectiveness of this strategy on four challenging benchmark datasets for fine-grained, object, scene, and texture classification: CUB-200-2011, PASCAL-VOC2007, MIT-Indoor, and DTD. Experiments using our proposed enhancement show promising results on all the datasets. In addition, our approach is capable of improving the performance of all generic CNN architectures.",
"Recently there has been a lot of work on pruning filters from deep convolutional neural networks (CNNs) with the intention of reducing computations. The key idea is to rank the filters based on a certain criterion (say, l1-norm, average percentage of zeros, etc) and retain only the top ranked filters. Once the low scoring filters are pruned away the remainder of the network is fine tuned and is shown to give performance comparable to the original unpruned network. In this work, we report experiments which suggest that the comparable performance of the pruned network is not due to the specific criterion chosen but due to the inherent plasticity of deep neural networks which allows them to recover from the loss of pruned filters once the rest of the filters are fine-tuned. Specifically, we show counter-intuitive results wherein by randomly pruning 25-50 filters from deep CNNs we are able to obtain the same performance as obtained by using state of the art pruning methods. We empirically validate our claims by doing an exhaustive evaluation with VGG-16 and ResNet-50. Further, we also evaluate a real world scenario where a CNN trained on all 1000 ImageNet classes needs to be tested on only a small set of classes at test time (say, only animals). We create a new benchmark dataset from ImageNet to evaluate such class specific pruning and show that even here a random pruning strategy gives close to state of the art performance. Lastly, unlike existing approaches which mainly focus on the task of image classification, in this work we also report results on object detection. We show that using a simple random pruning strategy we can achieve significant speed up in object detection (74 improvement in fps) while retaining the same accuracy as that of the original Faster RCNN model.",
"This paper proposes a novel approach to person re-identification, a fundamental task in distributed multi-camera surveillance systems. Although a variety of powerful algorithms have been presented in the past few years, most of them usually focus on designing hand-crafted features and learning metrics either individually or sequentially. Different from previous works, we formulate a unified deep ranking framework that jointly tackles both of these key components to maximize their strengths. We start from the principle that the correct match of the probe image should be positioned in the top rank within the whole gallery set. An effective learning-to-rank algorithm is proposed to minimize the cost corresponding to the ranking disorders of the gallery. The ranking model is solved with a deep convolutional neural network (CNN) that builds the relation between input image pairs and their similarity scores through joint representation learning directly from raw image pixels. The proposed framework allows us to get rid of feature engineering and does not rely on any assumption. An extensive comparative evaluation is given, demonstrating that our approach significantly outperforms all the state-of-the-art approaches, including both traditional and CNN-based methods on the challenging VIPeR, CUHK-01, and CAVIAR4REID datasets. In addition, our approach has better ability to generalize across datasets without fine-tuning."
]
} |
1901.09237 | 2950847180 | Digitally retouching images has become a popular trend, with people posting altered images on social media and even magazines posting flawless facial images of celebrities. Further, with advancements in Generative Adversarial Networks (GANs), now changing attributes and retouching have become very easy. Such synthetic alterations have adverse effect on face recognition algorithms. While researchers have proposed to detect image tampering, detecting GANs generated images has still not been explored. This paper proposes a supervised deep learning algorithm using Convolutional Neural Networks (CNNs) to detect synthetically altered images. The algorithm yields an accuracy of 99.65 on detecting retouching on the ND-IIITD dataset. It outperforms the previous state of the art which reported an accuracy of 87 on the database. For distinguishing between real images and images generated using GANs, the proposed algorithm yields an accuracy of 99.83 . | Retouching, makeup detection, face spoofing and morphing are widely studied areas, that can be considered similar to retouching detection. Recent work by @cite_2 makes use of supervised deep Boltzmann machine algorithm for detecting retouching on the ND-IIITD database. It also introduces the ND-IIITD dataset which consists of 2600 original and 2275 retouched images. It uses different facial parts to learn features for classification. In 2017, @cite_14 proposed an algorithm which uses semi-supervised autoencoders. The paper has reported results on the Multi-Demographic Retouched Faces (MDRF) dataset. Earlier research by Kee and Farid @cite_12 learned a support vector regression (SVR) between the retouched and original images. They used both geometric and photometric features for training the SVR on various celebrity images. | {
"cite_N": [
"@cite_14",
"@cite_12",
"@cite_2"
],
"mid": [
"2911434503",
"2432677034",
"2757024555"
],
"abstract": [
"Digitally retouching images has become a popular trend, with people posting altered images on social media and even magazines posting flawless facial images of celebrities. Further, with advancements in Generative Adversarial Networks (GANs), now changing attributes and retouching have become very easy. Such synthetic alterations have adverse effect on face recognition algorithms. While researchers have proposed to detect image tampering, detecting GANs generated images has still not been explored. This paper proposes a supervised deep learning algorithm using Convolutional Neural Networks (CNNs) to detect synthetically altered images. The algorithm yields an accuracy of 99.65 on detecting retouching on the ND-IIITD dataset. It outperforms the previous state of the art which reported an accuracy of 87 on the database. For distinguishing between real images and images generated using GANs, the proposed algorithm yields an accuracy of 99.83 .",
"Digitally altering, or retouching, face images is a common practice for images on social media, photo sharing websites, and even identification cards when the standards are not strictly enforced. This research demonstrates the effect of digital alterations on the performance of automatic face recognition, and also introduces an algorithm to classify face images as original or retouched with high accuracy. We first introduce two face image databases with unaltered and retouched images. Face recognition experiments performed on these databases show that when a retouched image is matched with its original image or an unaltered gallery image, the identification performance is considerably degraded, with a drop in matching accuracy of up to 25 . However, when images are retouched with the same style, the matching accuracy can be misleadingly high in comparison with matching original images. To detect retouching in face images, a novel supervised deep Boltzmann machine algorithm is proposed. It uses facial parts to learn discriminative features to classify face images as original or retouched. The proposed approach for classifying images as original or retouched yields an accuracy of over 87 on the data sets introduced in this paper and over 99 on three other makeup data sets used by previous researchers. This is a substantial increase in accuracy over the previous state-of-the-art algorithm, which has shown <50 accuracy in classifying original and retouched images from the ND-IIITD retouched faces database.",
"Digital retouching of face images is becoming more widespread due to the introduction of software packages that automate the task. Several researchers have introduced algorithms to detect whether a face image is original or retouched. However, previous work on this topic has not considered whether or how accuracy of retouching detection varies with the demography of face images. In this paper, we introduce a new Multi-Demographic Retouched Faces (MDRF) dataset, which contains images belonging to two genders, male and female, and three ethnicities, Indian, Chinese, and Caucasian. Further, retouched images are created using two different retouching software packages. The second major contribution of this research is a novel semi-supervised autoencoder incorporating \"subclass\" information to improve classification. The proposed approach outperforms existing state-of-the-art detection algorithms for the task of generalized retouching detection. Experiments conducted with multiple combinations of ethnicities show that accuracy of retouching detection can vary greatly based on the demographics of the training and testing images."
]
} |
1901.09244 | 2912003633 | Video recognition models have progressed significantly over the past few years, evolving from shallow classifiers trained on hand-crafted features to deep spatiotemporal networks. However, labeled video data required to train such models have not been able to keep up with the ever-increasing depth and sophistication of these networks. In this work, we propose an alternative approach to learning video representations that require no semantically labeled videos and instead leverages the years of effort in collecting and labeling large and clean still-image datasets. We do so by using state-of-the-art models pre-trained on image datasets as "teachers" to train video models in a distillation framework. We demonstrate that our method learns truly spatiotemporal features, despite being trained only using supervision from still-image networks. Moreover, it learns good representations across different input modalities, using completely uncurated raw video data sources and with different 2D teacher models. Our method obtains strong transfer performance, outperforming standard techniques for bootstrapping video architectures with image-based models by 16 . We believe that our approach opens up new approaches for learning spatiotemporal representations from unlabeled video data. | Video understanding, specifically for the task of human action recognition, is a well studied problem in computer vision. Analogously to the progress of image-based recognition methods, which have advanced from hand-crafted features @cite_14 @cite_34 to modern deep networks @cite_2 @cite_11 @cite_32 , video understanding methods have also evolved from hand-designed models @cite_19 @cite_13 @cite_37 to deep spatiotemporal networks @cite_26 @cite_33 . However, while image based recognition has seen dramatic gains in accuracy, improvements in video analysis have been more modest. In the still-image domain, deep models have greatly benefited from the availability of well-labeled datasets, such as ImageNet @cite_12 or Places @cite_31 . | {
"cite_N": [
"@cite_37",
"@cite_14",
"@cite_26",
"@cite_33",
"@cite_32",
"@cite_19",
"@cite_2",
"@cite_31",
"@cite_34",
"@cite_13",
"@cite_12",
"@cite_11"
],
"mid": [
"2176302750",
"2962778061",
"2511475724",
"2963218601",
"2048931426",
"2079169833",
"1932571874",
"110427074",
"2810685774",
"2010267071",
"2950971447",
"2136853139"
],
"abstract": [
"The recognition of human activities is one of the key problems in video understanding. Action recognition is challenging even for specific categories of videos, such as sports, that contain only a small set of actions. Interestingly, sports videos are accompanied by detailed commentaries available online, which could be used to perform action annotation in a weakly-supervised setting. For the specific case of Cricket videos, we address the challenge of temporal segmentation and annotation of ctions with semantic descriptions. Our solution consists of two stages. In the first stage, the video is segmented into \"scenes\", by utilizing the scene category information extracted from text-commentary. The second stage consists of classifying video-shots as well as the phrases in the textual description into various categories. The relevant phrases are then suitably mapped to the video-shots. The novel aspect of this work is the fine temporal scale at which semantic information is assigned to the video. As a result of our approach, we enable retrieval of specific actions that last only a few seconds, from several hours of video. This solution yields a large number of labeled exemplars, with no manual effort, that could be used by machine learning algorithms to learn complex actions.",
"Deep learning for human action recognition in videos is making significant progress, but is slowed down by its dependency on expensive manual labeling of large video collections. In this work, we investigate the generation of synthetic training data for action recognition, as it has recently shown promising results for a variety of other computer vision tasks. We propose an interpretable parametric generative model of human action videos that relies on procedural generation and other computer graphics techniques of modern game engines. We generate a diverse, realistic, and physically plausible dataset of human action videos, called PHAV for Procedural Human Action Videos. It contains a total of 39,982 videos, with more than 1,000 examples for each action of 35 categories. Our approach is not limited to existing motion capture sequences, and we procedurally define 14 synthetic actions. We introduce a deep multi-task representation learning architecture to mix synthetic and real videos, even if the action categories differ. Our experiments on the UCF101 and HMDB51 benchmarks suggest that combining our large set of synthetic videos with small real-world datasets can boost recognition performance, significantly outperforming fine-tuning state-of-the-art unsupervised generative models of videos.",
"Action recognition in videos is a challenging task due to the complexity of the spatio-temporal patterns to model and the difficulty to acquire and learn on large quantities of video data. Deep learning, although a breakthrough for image classification and showing promise for videos, has still not clearly superseded action recognition methods using hand-crafted features, even when training on massive datasets. In this paper, we introduce hybrid video classification architectures based on carefully designed unsupervised representations of hand-crafted spatio-temporal features classified by supervised deep networks. As we show in our experiments on five popular benchmarks for action recognition, our hybrid model combines the best of both worlds: it is data efficient (trained on 150 to 10000 short clips) and yet improves significantly on the state of the art, including recent deep models trained on millions of manually labelled images and videos.",
"Understanding human actions in visual data is tied to advances in complementary research areas including object recognition, human dynamics, domain adaptation and semantic segmentation. Over the last decade, human action analysis evolved from earlier schemes that are often limited to controlled environments to nowadays advanced solutions that can learn from millions of videos and apply to almost all daily activities. Given the broad range of applications from video surveillance to humancomputer interaction, scientific milestones in action recognition are achieved more rapidly, eventually leading to the demise of what used to be good in a short time. This motivated us to provide a comprehensive review of the notable steps taken towards recognizing human actions. To this end, we start our discussion with the pioneering methods that use handcrafted representations, and then, navigate into the realm of deep learning based approaches. We aim to remain objective throughout this survey, touching upon encouraging improvements as well as inevitable fallbacks, in the hope of raising fresh questions and motivating new research directions for the reader. We provide a detailed review of the work on human action recognition over the past decade.We refer to actions as meaningful human motions.Including Hand-crafted representations methods, we review the impact of Deep-nets on action recognition.We follow a systematic taxonomy to highlight the essence of both Hand-crafted and Deep-net solutions.We present a comparison of methods at their algorithmic level and performance.",
"Human action recognition in video is important in many computer vision applications such as automated surveillance. Human actions can be compactly encoded using a sparse set of local spatio-temporal salient features at different scales. The existing bottom-up methods construct a single dictionary of action primitives from the joint features of all scales and hence, a single action representation. This representation cannot fully exploit the complementary characteristics of the motions across different scales. To address this problem, we introduce the concept of learning multiple dictionaries of action primitives at different resolutions and consequently, multiple scale-specific representations for a given video sample. Using a decoupled fusion of multiple representations, we improved the human classification accuracy of realistic benchmark databases by about 5 , compared with the state-of-the art methods.",
"This paper presents an approach for human action recognition by finding the discriminative key frames from a video sequence and representing them with the distribution of local motion features and their spatiotemporal arrangements. In this approach, the key frames of the video sequence are selected by their discriminative power and represented by the local motion features detected in them and integrated from their temporal neighbors. In the key frame’s representation, the spatial arrangements of the motion features are captured in a hierarchical spatial pyramid structure. By using frame by frame voting for the recognition, experiments have demonstrated improved performances over most of the other known methods on the popular benchmark data sets. Recognizinghumanactionfromimagesequencesis an appealingyet challengingproblem in computer vision with many applications including motion capture, human-computer interaction, environment control, and security surveillance. In this paper, we focus on recognizing the activities of a person in an image sequence from local motion features and their spatiotemporal arrangements. Our approach is motivated by the recent success of “bag-of-words” model for general object recognition in computer vision[21, 14]. This representation, which is adapted from the text retrieval literature, models the object by the distribution of words from a fixed visual code book, which is usually obtained by vector quantization of local image visual features. However, this method discards the spatial and the temporal relations among these visual features, which could be helpful in object recognition. Addressing this problem, our approach uses a hierarchical representation for the key frames of a given video sequence to integrate information from both the spatial and the temporal domains. We first apply a spatiotemporal feature detector to the video sequence and obtain the local motion features. Then we generate a visual word code book by quantization of the local motion features and assign a word label to each of them. Next we select key frames of the video sequence by their discriminative power. Then, for each key frame, we integrate the visual words from its nearby frames, divide the key frame spatially into finer subdivisions and compute in each cell the histograms of the visual words detected in this key frame and its temporal neighbors. Finally, we concatenate the histograms from all cells and use",
"Whereas the action recognition problem has become a hot topic within computer vision, the detection of fights or in general aggressive behavior has been comparatively less studied. Such capability may be extremely useful in some video surveillance scenarios like in prisons, psychiatric centers or even embedded in camera phones. Recent work has considered the well-known Bag-of-Words framework often used in generic action recognition for the specific problem of fight detection. Under this framework, spatio-temporal features are extracted from the video sequences and used for classification. Despite encouraging results in which near 90 accuracy rates were achieved for this specific task, the computational cost of extracting such features is prohibitive for practical applications, particularly in surveillance and media rating systems. The task of violence detection may have, however, specific features that can be leveraged. Inspired by psychology results that suggest that kinematic features alone are discriminant for specific actions, this work proposes a novel method which uses extreme acceleration patterns as the main feature. These extreme accelerations are efficiently estimated by applying the Radon transform to the power spectrum of consecutive frames. Experiments show that accuracy improvements of up to 12 are achieved with respect to state-of-the-art generic action recognition methods. Most importantly, the proposed method is at least 15 times faster.",
"Human action is, in general, characterized by a sequence of specific body postures. An action is often recognizable from a single view of an action specific posture. In this paper we demonstrate that specific actions can be recognized in long video sequence by matching shape information extracted from individual frames to stored prototypes representing key frames of the action. The matching algorithm is tolerant to substantial deformation between image and prototype and recognizes qualitatively similar image shapes produced by the body postures. The algorithm is applied to the recognition of specific tennis strokes in up to 2 min long video sequences. Using key-frames within the sequence all (1015) forehand strokes are found with no false positives. The backhand strokes are detected with somewhat lower performance. We conclude that this type of approach to action recognition has large potential for application to automatic video analysis and editing.",
"Derived from rapid advances in computer vision and machine learning, video analysis tasks have been moving from inferring the present state to predicting the future state. Vision-based action recognition and prediction from videos are such tasks, where action recognition is to infer human actions (present state) based upon complete action executions, and action prediction to predict human actions (future state) based upon incomplete action executions. These two tasks have become particularly prevalent topics recently because of their explosively emerging real-world applications, such as visual surveillance, autonomous driving vehicle, entertainment, and video retrieval, etc. Many attempts have been devoted in the last a few decades in order to build a robust and effective framework for action recognition and prediction. In this paper, we survey the complete state-of-the-art techniques in the action recognition and prediction. Existing models, popular algorithms, technical difficulties, popular action databases, evaluation protocols, and promising future directions are also provided with systematic discussions.",
"Action recognition is an important problem in computer vision and has received substantial attention in recent years. However, it remains very challenging due to the complex interaction of static and dynamic information, as well as the high computational cost of processing video data. This paper aims to apply the success of static image semantic recognition to the video domain, by leveraging both static and motion based descriptors in different stages of the semantic ladder. We examine the effects of three types of features: low-level dynamic descriptors, intermediate-level static deep architecture outputs, and static high-level semantics. In order to combine such heterogeneous sources of information, we employ a scalable method to fuse these features. Through extensive experimental evaluations, we demonstrate that the proposed framework significantly improves action classification performance. We have obtained an accuracy of 89.59 and 62.88 on the well-known UCF-101 and HMDB-51 benchmarks, respectively, which compare favorably with the state-of-the-art.",
"Deep convolutional networks have achieved great success for visual recognition in still images. However, for action recognition in videos, the advantage over traditional methods is not so evident. This paper aims to discover the principles to design effective ConvNet architectures for action recognition in videos and learn these models given limited training samples. Our first contribution is temporal segment network (TSN), a novel framework for video-based action recognition. which is based on the idea of long-range temporal structure modeling. It combines a sparse temporal sampling strategy and video-level supervision to enable efficient and effective learning using the whole action video. The other contribution is our study on a series of good practices in learning ConvNets on video data with the help of temporal segment network. Our approach obtains the state-the-of-art performance on the datasets of HMDB51 ( @math ) and UCF101 ( @math ). We also visualize the learned ConvNet models, which qualitatively demonstrates the effectiveness of temporal segment network and the proposed good practices.",
"Visual recognition of human actions in video clips has been an active field of research in recent years. However, most published methods either analyse an entire video and assign it a single action label, or use relatively large look-ahead to classify each frame. Contrary to these strategies, human vision proves that simple actions can be recognised almost instantaneously. In this paper, we present a system for action recognition from very short sequences (ldquosnippetsrdquo) of 1-10 frames, and systematically evaluate it on standard data sets. It turns out that even local shape and optic flow for a single frame are enough to achieve ap90 correct recognitions, and snippets of 5-7 frames (0.3-0.5 seconds of video) are enough to achieve a performance similar to the one obtainable with the entire video sequence."
]
} |
1901.09244 | 2912003633 | Video recognition models have progressed significantly over the past few years, evolving from shallow classifiers trained on hand-crafted features to deep spatiotemporal networks. However, labeled video data required to train such models have not been able to keep up with the ever-increasing depth and sophistication of these networks. In this work, we propose an alternative approach to learning video representations that require no semantically labeled videos and instead leverages the years of effort in collecting and labeling large and clean still-image datasets. We do so by using state-of-the-art models pre-trained on image datasets as "teachers" to train video models in a distillation framework. We demonstrate that our method learns truly spatiotemporal features, despite being trained only using supervision from still-image networks. Moreover, it learns good representations across different input modalities, using completely uncurated raw video data sources and with different 2D teacher models. Our method obtains strong transfer performance, outperforming standard techniques for bootstrapping video architectures with image-based models by 16 . We believe that our approach opens up new approaches for learning spatiotemporal representations from unlabeled video data. | Until recently, video datasets have either been well-labeled but small @cite_47 @cite_24 @cite_7 , or large but weakly-labeled @cite_44 @cite_22 . A recently introduced dataset, Kinetics @cite_35 , is currently the largest well-annotated dataset, with around 300K videos labeled into 400 categories (we note a larger version with 600K videos in 600 categories was recently released). It is nearly two orders of magnitude larger than previously established benchmarks in video classification @cite_47 @cite_24 . As expected, pre-training networks on this dataset has yielded significant gains in accuracy @cite_39 on many standard benchmarks @cite_47 @cite_24 @cite_7 , and have won CVPR 2017 ActivityNet and Charades challenges. However, it is worth noting that this dataset was collected at a significant curation and annotation effort @cite_35 . | {
"cite_N": [
"@cite_35",
"@cite_22",
"@cite_7",
"@cite_39",
"@cite_24",
"@cite_44",
"@cite_47"
],
"mid": [
"2655063496",
"1994694952",
"2524365899",
"2337252826",
"2108710284",
"2949594863",
"1777628566"
],
"abstract": [
"The YouTube-8M video classification challenge requires teams to classify 0.7 million videos into one or more of 4,716 classes. In this Kaggle competition, we placed in the top 3 out of 650 participants using released video and audio features. Beyond that, we extend the original competition by including text information in the classification, making this a truly multi-modal approach with vision, audio and text. The newly introduced text data is termed as YouTube-8M-Text. We present a classification framework for the joint use of text, visual and audio features, and conduct an extensive set of experiments to quantify the benefit that this additional mode brings. The inclusion of text yields state-of-the-art results, e.g. 86.7 GAP on the YouTube-8M-Text validation dataset.",
"Automatic categorization of videos in a Web-scale unconstrained collection such as YouTube is a challenging task. A key issue is how to build an effective training set in the presence of missing, sparse or noisy labels. We propose to achieve this by first manually creating a small labeled set and then extending it using additional sources such as related videos, searched videos, and text-based webpages. The data from such disparate sources has different properties and labeling quality, and thus fusing them in a coherent fashion is another practical challenge. We propose a fusion framework in which each data source is first combined with the manually-labeled set independently. Then, using the hierarchical taxonomy of the categories, a Conditional Random Field (CRF) based fusion strategy is designed. Based on the final fused classifier, category labels are predicted for the new videos. Extensive experiments on about 80K videos from 29 most frequent categories in YouTube show the effectiveness of the proposed method for categorizing large-scale wild Web videos1.",
"Many recent advancements in Computer Vision are attributed to large datasets. Open-source software packages for Machine Learning and inexpensive commodity hardware have reduced the barrier of entry for exploring novel approaches at scale. It is possible to train models over millions of examples within a few days. Although large-scale datasets exist for image understanding, such as ImageNet, there are no comparable size video classification datasets. In this paper, we introduce YouTube-8M, the largest multi-label video classification dataset, composed of 8 million videos (500K hours of video), annotated with a vocabulary of 4800 visual entities. To get the videos and their labels, we used a YouTube video annotation system, which labels videos with their main topics. While the labels are machine-generated, they have high-precision and are derived from a variety of human-based signals including metadata and query click signals. We filtered the video labels (Knowledge Graph entities) using both automated and manual curation strategies, including asking human raters if the labels are visually recognizable. Then, we decoded each video at one-frame-per-second, and used a Deep CNN pre-trained on ImageNet to extract the hidden representation immediately prior to the classification layer. Finally, we compressed the frame features and make both the features and video-level labels available for download. We trained various (modest) classification models on the dataset, evaluated them using popular evaluation metrics, and report them as baselines. Despite the size of the dataset, some of our models train to convergence in less than a day on a single machine using TensorFlow. We plan to release code for training a TensorFlow model and for computing metrics.",
"Computer vision has a great potential to help our daily lives by searching for lost keys, watering flowers or reminding us to take a pill. To succeed with such tasks, computer vision methods need to be trained from real and diverse examples of our daily dynamic scenes. While most of such scenes are not particularly exciting, they typically do not appear on YouTube, in movies or TV broadcasts. So how do we collect sufficiently many diverse but boring samples representing our lives? We propose a novel Hollywood in Homes approach to collect such data. Instead of shooting videos in the lab, we ensure diversity by distributing and crowdsourcing the whole process of video creation from script writing to video recording and annotation. Following this procedure we collect a new dataset, Charades, with hundreds of people recording videos in their own homes, acting out casual everyday activities. The dataset is composed of 9,848 annotated videos with an average length of 30 s, showing activities of 267 people from three continents. Each video is annotated by multiple free-text descriptions, action labels, action intervals and classes of interacted objects. In total, Charades provides 27,847 video descriptions, 66,500 temporally localized intervals for 157 action classes and 41,104 labels for 46 object classes. Using this rich data, we evaluate and provide baseline results for several tasks including action recognition and automatic description generation. We believe that the realism, diversity, and casual nature of this dataset will present unique challenges and new opportunities for computer vision community.",
"We are given a set of video clips, each one annotated with an ordered list of actions, such as “walk” then “sit” then “answer phone” extracted from, for example, the associated text script. We seek to temporally localize the individual actions in each clip as well as to learn a discriminative classifier for each action. We formulate the problem as a weakly supervised temporal assignment with ordering constraints. Each video clip is divided into small time intervals and each time interval of each video clip is assigned one action label, while respecting the order in which the action labels appear in the given annotations. We show that the action label assignment can be determined together with learning a classifier for each action in a discriminative manner. We evaluate the proposed model on a new and challenging dataset of 937 video clips with a total of 787720 frames containing sequences of 16 different actions from 69 Hollywood movies.",
"We are given a set of video clips, each one annotated with an ordered list of actions, such as \"walk\" then \"sit\" then \"answer phone\" extracted from, for example, the associated text script. We seek to temporally localize the individual actions in each clip as well as to learn a discriminative classifier for each action. We formulate the problem as a weakly supervised temporal assignment with ordering constraints. Each video clip is divided into small time intervals and each time interval of each video clip is assigned one action label, while respecting the order in which the action labels appear in the given annotations. We show that the action label assignment can be determined together with learning a classifier for each action in a discriminative manner. We evaluate the proposed model on a new and challenging dataset of 937 video clips with a total of 787720 frames containing sequences of 16 different actions from 69 Hollywood movies.",
"In this paper, we study the challenging problem of categorizing videos according to high-level semantics such as the existence of a particular human action or a complex event. Although extensive efforts have been devoted in recent years, most existing works combined multiple video features using simple fusion strategies and neglected the utilization of inter-class semantic relationships. This paper proposes a novel unified framework that jointly exploits the feature relationships and the class relationships for improved categorization performance. Specifically, these two types of relationships are estimated and utilized by imposing regularizations in the learning process of a deep neural network (DNN). Through arming the DNN with better capability of harnessing both the feature and the class relationships, the proposed regularized DNN (rDNN) is more suitable for modeling video semantics. We show that rDNN produces better performance over several state-of-the-art approaches. Competitive results are reported on the well-known Hollywood2 and Columbia Consumer Video benchmarks. In addition, to stimulate future research on large scale video categorization, we collect and release a new benchmark dataset, called FCVID, which contains 91,223 Internet videos and 239 manually annotated categories."
]
} |
1901.09244 | 2912003633 | Video recognition models have progressed significantly over the past few years, evolving from shallow classifiers trained on hand-crafted features to deep spatiotemporal networks. However, labeled video data required to train such models have not been able to keep up with the ever-increasing depth and sophistication of these networks. In this work, we propose an alternative approach to learning video representations that require no semantically labeled videos and instead leverages the years of effort in collecting and labeling large and clean still-image datasets. We do so by using state-of-the-art models pre-trained on image datasets as "teachers" to train video models in a distillation framework. We demonstrate that our method learns truly spatiotemporal features, despite being trained only using supervision from still-image networks. Moreover, it learns good representations across different input modalities, using completely uncurated raw video data sources and with different 2D teacher models. Our method obtains strong transfer performance, outperforming standard techniques for bootstrapping video architectures with image-based models by 16 . We believe that our approach opens up new approaches for learning spatiotemporal representations from unlabeled video data. | The challenge in generating large-scale well-labeled video datasets stems from the fact that a human annotator has to spend much longer to label a video compared to a single image. Previous work has attempted to reduce this labeling effort through heuristics @cite_8 , but these methods still require a human annotator to clean up the final labels. There has also been some work in learning unsupervised video representations @cite_15 @cite_21 , however has typically lead to inferior results compared to supervised features. | {
"cite_N": [
"@cite_15",
"@cite_21",
"@cite_8"
],
"mid": [
"2524365899",
"2949643062",
"2108710284"
],
"abstract": [
"Many recent advancements in Computer Vision are attributed to large datasets. Open-source software packages for Machine Learning and inexpensive commodity hardware have reduced the barrier of entry for exploring novel approaches at scale. It is possible to train models over millions of examples within a few days. Although large-scale datasets exist for image understanding, such as ImageNet, there are no comparable size video classification datasets. In this paper, we introduce YouTube-8M, the largest multi-label video classification dataset, composed of 8 million videos (500K hours of video), annotated with a vocabulary of 4800 visual entities. To get the videos and their labels, we used a YouTube video annotation system, which labels videos with their main topics. While the labels are machine-generated, they have high-precision and are derived from a variety of human-based signals including metadata and query click signals. We filtered the video labels (Knowledge Graph entities) using both automated and manual curation strategies, including asking human raters if the labels are visually recognizable. Then, we decoded each video at one-frame-per-second, and used a Deep CNN pre-trained on ImageNet to extract the hidden representation immediately prior to the classification layer. Finally, we compressed the frame features and make both the features and video-level labels available for download. We trained various (modest) classification models on the dataset, evaluated them using popular evaluation metrics, and report them as baselines. Despite the size of the dataset, some of our models train to convergence in less than a day on a single machine using TensorFlow. We plan to release code for training a TensorFlow model and for computing metrics.",
"Large-scale annotated datasets allow AI systems to learn from and build upon the knowledge of the crowd. Many crowdsourcing techniques have been developed for collecting image annotations. These techniques often implicitly rely on the fact that a new input image takes a negligible amount of time to perceive. In contrast, we investigate and determine the most cost-effective way of obtaining high-quality multi-label annotations for temporal data such as videos. Watching even a short 30-second video clip requires a significant time investment from a crowd worker; thus, requesting multiple annotations following a single viewing is an important cost-saving strategy. But how many questions should we ask per video? We conclude that the optimal strategy is to ask as many questions as possible in a HIT (up to 52 binary questions after watching a 30-second video clip in our experiments). We demonstrate that while workers may not correctly answer all questions, the cost-benefit analysis nevertheless favors consensus from multiple such cheap-yet-imperfect iterations over more complex alternatives. When compared with a one-question-per-video baseline, our method is able to achieve a 10 improvement in recall (76.7 ours versus 66.7 baseline) at comparable precision (83.8 ours versus 83.0 baseline) in about half the annotation time (3.8 minutes ours compared to 7.1 minutes baseline). We demonstrate the effectiveness of our method by collecting multi-label annotations of 157 human activities on 1,815 videos.",
"We are given a set of video clips, each one annotated with an ordered list of actions, such as “walk” then “sit” then “answer phone” extracted from, for example, the associated text script. We seek to temporally localize the individual actions in each clip as well as to learn a discriminative classifier for each action. We formulate the problem as a weakly supervised temporal assignment with ordering constraints. Each video clip is divided into small time intervals and each time interval of each video clip is assigned one action label, while respecting the order in which the action labels appear in the given annotations. We show that the action label assignment can be determined together with learning a classifier for each action in a discriminative manner. We evaluate the proposed model on a new and challenging dataset of 937 video clips with a total of 787720 frames containing sequences of 16 different actions from 69 Hollywood movies."
]
} |
1901.09244 | 2912003633 | Video recognition models have progressed significantly over the past few years, evolving from shallow classifiers trained on hand-crafted features to deep spatiotemporal networks. However, labeled video data required to train such models have not been able to keep up with the ever-increasing depth and sophistication of these networks. In this work, we propose an alternative approach to learning video representations that require no semantically labeled videos and instead leverages the years of effort in collecting and labeling large and clean still-image datasets. We do so by using state-of-the-art models pre-trained on image datasets as "teachers" to train video models in a distillation framework. We demonstrate that our method learns truly spatiotemporal features, despite being trained only using supervision from still-image networks. Moreover, it learns good representations across different input modalities, using completely uncurated raw video data sources and with different 2D teacher models. Our method obtains strong transfer performance, outperforming standard techniques for bootstrapping video architectures with image-based models by 16 . We believe that our approach opens up new approaches for learning spatiotemporal representations from unlabeled video data. | The question we pose is: since labeling images is faster, and since we already have large, well-labeled image datasets such as ImageNet, can we instead use these to bootstrap the learning of spatiotemporal video architectures? Unsurprisingly, various previous approaches have attempted this. The popular two-stream architecture @cite_33 uses individual frames from the video as input. Hence it initializes the RGB stream of the network with weights pre-trained on ImageNet and then fine-tunes them for action classification on the action dataset. More recent variants of two-stream architectures have also initialized the flow stream @cite_28 from weights pretrained on ImageNet by viewing optical flow as a grayscale image. | {
"cite_N": [
"@cite_28",
"@cite_33"
],
"mid": [
"2156303437",
"2952186347"
],
"abstract": [
"We investigate architectures of discriminatively trained deep Convolutional Networks (ConvNets) for action recognition in video. The challenge is to capture the complementary information on appearance from still frames and motion between frames. We also aim to generalise the best performing hand-crafted features within a data-driven learning framework. Our contribution is three-fold. First, we propose a two-stream ConvNet architecture which incorporates spatial and temporal networks. Second, we demonstrate that a ConvNet trained on multi-frame dense optical flow is able to achieve very good performance in spite of limited training data. Finally, we show that multitask learning, applied to two different action classification datasets, can be used to increase the amount of training data and improve the performance on both. Our architecture is trained and evaluated on the standard video actions benchmarks of UCF-101 and HMDB-51, where it is competitive with the state of the art. It also exceeds by a large margin previous attempts to use deep nets for video classification.",
"We investigate architectures of discriminatively trained deep Convolutional Networks (ConvNets) for action recognition in video. The challenge is to capture the complementary information on appearance from still frames and motion between frames. We also aim to generalise the best performing hand-crafted features within a data-driven learning framework. Our contribution is three-fold. First, we propose a two-stream ConvNet architecture which incorporates spatial and temporal networks. Second, we demonstrate that a ConvNet trained on multi-frame dense optical flow is able to achieve very good performance in spite of limited training data. Finally, we show that multi-task learning, applied to two different action classification datasets, can be used to increase the amount of training data and improve the performance on both. Our architecture is trained and evaluated on the standard video actions benchmarks of UCF-101 and HMDB-51, where it is competitive with the state of the art. It also exceeds by a large margin previous attempts to use deep nets for video classification."
]
} |
1901.09244 | 2912003633 | Video recognition models have progressed significantly over the past few years, evolving from shallow classifiers trained on hand-crafted features to deep spatiotemporal networks. However, labeled video data required to train such models have not been able to keep up with the ever-increasing depth and sophistication of these networks. In this work, we propose an alternative approach to learning video representations that require no semantically labeled videos and instead leverages the years of effort in collecting and labeling large and clean still-image datasets. We do so by using state-of-the-art models pre-trained on image datasets as "teachers" to train video models in a distillation framework. We demonstrate that our method learns truly spatiotemporal features, despite being trained only using supervision from still-image networks. Moreover, it learns good representations across different input modalities, using completely uncurated raw video data sources and with different 2D teacher models. Our method obtains strong transfer performance, outperforming standard techniques for bootstrapping video architectures with image-based models by 16 . We believe that our approach opens up new approaches for learning spatiotemporal representations from unlabeled video data. | However, such initializations are only applicable to video models that use 2D convolutions, analogous to those applied in CNNs for still-images. What about more complex, truly spatiotemporal models, such as 3D convolutional architectures @cite_26 ? Until recently, such models have largely been limited to pre-training on large but weakly-labeled video datasets, such as Sports1M @cite_44 . Recent work @cite_39 @cite_29 proposed a nice alternative, consisting of inflating standard 2D CNNs kernels to 3D, by simply replicating the 2D kernels in time. While effective in getting strong performance on large benchmarks, on small datasets this approach tends to bias video models to be close to static replicas of the image models. Moreover, such initialization constrains the 3D architecture to be identical to the 2D CNN, except for the additional third dimension in kernels. This effectively restricts the design of video models to extensions of what works best in the still-image domain, which may not be the architectures for video analysis. | {
"cite_N": [
"@cite_44",
"@cite_29",
"@cite_26",
"@cite_39"
],
"mid": [
"2748434587",
"2761659801",
"2963616706",
"2883429621"
],
"abstract": [
"Convolutional neural networks with spatio-temporal 3D kernels (3D CNNs) have an ability to directly extract spatio-temporal features from videos for action recognition. Although the 3D kernels tend to overfit because of a large number of their parameters, the 3D CNNs are greatly improved by using recent huge video databases. However, the architecture of 3D CNNs is relatively shallow against to the success of very deep neural networks in 2D-based CNNs, such as residual networks (ResNets). In this paper, we propose a 3D CNNs based on ResNets toward a better action representation. We describe the training procedure of our 3D ResNets in details. We experimentally evaluate the 3D ResNets on the ActivityNet and Kinetics datasets. The 3D ResNets trained on the Kinetics did not suffer from overfitting despite the large number of parameters of the model, and achieved better performance than relatively shallow networks, such as C3D. Our code and pretrained models (e.g. Kinetics and ActivityNet) are publicly available at this https URL",
"Convolutional Neural Networks (CNN) have been regarded as a powerful class of models for image recognition problems. Nevertheless, it is not trivial when utilizing a CNN for learning spatio-temporal video representation. A few studies have shown that performing 3D convolutions is a rewarding approach to capture both spatial and temporal dimensions in videos. However, the development of a very deep 3D CNN from scratch results in expensive computational cost and memory demand. A valid question is why not recycle off-the-shelf 2D networks for a 3D CNN. In this paper, we devise multiple variants of bottleneck building blocks in a residual learning framework by simulating @math convolutions with @math convolutional filters on spatial domain (equivalent to 2D CNN) plus @math convolutions to construct temporal connections on adjacent feature maps in time. Furthermore, we propose a new architecture, named Pseudo-3D Residual Net (P3D ResNet), that exploits all the variants of blocks but composes each in different placement of ResNet, following the philosophy that enhancing structural diversity with going deep could improve the power of neural networks. Our P3D ResNet achieves clear improvements on Sports-1M video classification dataset against 3D CNN and frame-based 2D CNN by 5.3 and 1.8 , respectively. We further examine the generalization performance of video representation produced by our pre-trained P3D ResNet on five different benchmarks and three different tasks, demonstrating superior performances over several state-of-the-art techniques.",
"Convolutional neural networks with spatio-temporal 3D kernels (3D CNNs) have an ability to directly extract spatiotemporal features from videos for action recognition. Although the 3D kernels tend to overfit because of a large number of their parameters, the 3D CNNs are greatly improved by using recent huge video databases. However, the architecture of3D CNNs is relatively shallow against to the success of very deep neural networks in 2D-based CNNs, such as residual networks (ResNets). In this paper, we propose a 3D CNNs based on ResNets toward a better action representation. We describe the training procedure of our 3D ResNets in details. We experimentally evaluate the 3D ResNets on the ActivityNet and Kinetics datasets. The 3D ResNets trained on the Kinetics did not suffer from overfitting despite the large number of parameters of the model, and achieved better performance than relatively shallow networks, such as C3D. Our code and pretrained models (e.g. Kinetics and ActivityNet) are publicly available at https: github.com kenshohara 3D-ResNets.",
"Despite the steady progress in video analysis led by the adoption of convolutional neural networks (CNNs), the relative improvement has been less drastic as that in 2D static image classification. Three main challenges exist including spatial (image) feature representation, temporal information representation, and model computation complexity. It was recently shown by Carreira and Zisserman that 3D CNNs, inflated from 2D networks and pretrained on ImageNet, could be a promising way for spatial and temporal representation learning. However, as for model computation complexity, 3D CNNs are much more expensive than 2D CNNs and prone to overfit. We seek a balance between speed and accuracy by building an effective and efficient video classification system through systematic exploration of critical network design choices. In particular, we show that it is possible to replace many of the 3D convolutions by low-cost 2D convolutions. Rather surprisingly, best result (in both speed and accuracy) is achieved when replacing the 3D convolutions at the bottom of the network, suggesting that temporal representation learning on high-level “semantic” features is more useful. Our conclusion generalizes to datasets with very different properties. When combined with several other cost-effective designs including separable spatial temporal convolution and feature gating, our system results in an effective video classification system that that produces very competitive results on several action classification benchmarks (Kinetics, Something-something, UCF101 and HMDB), as well as two action detection (localization) benchmarks (JHMDB and UCF101-24)."
]
} |
1907.09597 | 2963065757 | In this paper we explore how actor-critic methods in deep reinforcement learning, in particular Asynchronous Advantage Actor-Critic (A3C), can be extended with agent modeling. Inspired by recent works on representation learning and multiagent deep reinforcement learning, we propose two architectures to perform agent modeling: the first one based on parameter sharing, and the second one based on agent policy features. Both architectures aim to learn other agents' policies as auxiliary tasks, besides the standard actor (policy) and critic (values). We performed experiments in both cooperative and competitive domains. The former is a problem of coordinated multiagent object transportation and the latter is a two-player mini version of the Pommerman game. Our results show that the proposed architectures stabilize learning and outperform the standard A3C architecture when learning a best response in terms of expected rewards. | Deep Reinforcement Opponent Network (DRON) @cite_5 was the first DRL work that performed opponent modeling. DRON's idea is to have two networks: one learns @math values (similar to DQN @cite_2 ) and a second learns a representation of the opponent's policy. DRON used hand-crafted features to define the opponent network. In contrast, Deep Policy Inference Q-Network (DPIQN) and Deep Policy Inference Recurrent Q-Network (DPIRQN) @cite_0 learned opponent directly from raw observations of the other agents. The way to learn these policy features is by means of auxiliary tasks @cite_19 that provide additional learning goals; in this case, the auxiliary task is to learn the opponent's policy. Then, the @math value function of the learning agent is conditioned on the policy features, which aim to reduce the non-stationarity of the multiagent environment. In contrast, our proposals do not need an experience replay buffer, learn completely on-policy and we make use of full parameter sharing @cite_23 . | {
"cite_N": [
"@cite_0",
"@cite_19",
"@cite_23",
"@cite_2",
"@cite_5"
],
"mid": [
"2963259091",
"2152342063",
"2603550564",
"2904455790",
"2121092017"
],
"abstract": [
"We present DPIQN, a deep policy inference Q-network that targets multi-agent systems composed of controllable agents, collaborators, and opponents that interact with each other. We focus on one challenging issue in such systems---modeling agents with varying strategies---and propose to employ \"policy features'' learned from raw observations (e.g., raw images) of collaborators and opponents by inferring their policies. DPIQN incorporates the learned policy features as a hidden vector into its own deep Q-network (DQN), such that it is able to predict better Q values for the controllable agents than the state-of-the-art deep reinforcement learning models. We further propose an enhanced version of DPIQN, called deep recurrent policy inference Q-network (DRPIQN), for handling partial observability. Both DPIQN and DRPIQN are trained by an adaptive training procedure, which adjusts the network's attention to learn the policy features and its own Q-values at different phases of the training process. We present a comprehensive analysis of DPIQN and DRPIQN, and highlight their effectiveness and generalizability in various multi-agent settings. Our models are evaluated in a classic soccer game involving both competitive and collaborative scenarios. Experimental results performed on 1 vs. 1 and 2 vs. 2 games show that DPIQN and DRPIQN demonstrate superior performance to the baseline DQN and deep recurrent Q-network (DRQN) models. We also explore scenarios in which collaborators or opponents dynamically change their policies, and show that DPIQN and DRPIQN do lead to better overall performance in terms of stability and mean scores.",
"We use single-agent and multi-agent Reinforcement Learning (RL) for learning dialogue policies in a resource allocation negotiation scenario. Two agents learn concurrently by interacting with each other without any need for simulated users (SUs) to train against or corpora to learn from. In particular, we compare the Qlearning, Policy Hill-Climbing (PHC) and Win or Learn Fast Policy Hill-Climbing (PHC-WoLF) algorithms, varying the scenario complexity (state space size), the number of training episodes, the learning rate, and the exploration rate. Our results show that generally Q-learning fails to converge whereas PHC and PHC-WoLF always converge and perform similarly. We also show that very high gradually decreasing exploration rates are required for convergence. We conclude that multiagent RL of dialogue policies is a promising alternative to using single-agent RL and SUs or learning directly from corpora.",
"Standard deep reinforcement learning methods such as Deep Q-Networks (DQN) for multiple tasks (domains) face scalability problems due to large search spaces. This paper proposes a three-stage method for multi-domain dialogue policy learning-termed NDQN, and applies it to an information-seeking spoken dialogue system in the domains of restaurants and hotels. In this method, the first stage does multi-policy learning via a network of DQN agents; the second makes use of compact state representations by compressing raw inputs; and the third stage applies a pre-training phase for bootstraping the behaviour of agents in the network. Experimental results comparing DQN (baseline) versus NDQN (proposed) using simulations report that the proposed method exhibits better scalability and is promising for optimising the behaviour of multi-domain dialogue systems. An additional evaluation reports that the NDQN agents outperformed a K-Nearest Neighbour baseline in task success and dialogue length, yielding more efficient and successful dialogues.",
"Despite the recent advances of deep reinforcement learning (DRL), agents trained by DRL tend to be brittle and sensitive to the training environment, especially in the multi-agent scenarios. In the multi-agent setting, a DRL agent’s policy can easily get stuck in a poor local optima w.r.t. its training partners – the learned policy may be only locally optimal to other agents’ current policies. In this paper, we focus on the problem of training robust DRL agents with continuous actions in the multi-agent learning setting so that the trained agents can still generalize when its opponents’ policies alter. To tackle this problem, we proposed a new algorithm, MiniMax Multi-agent Deep Deterministic Policy Gradient (M3DDPG) with the following contributions: (1) we introduce a minimax extension of the popular multi-agent deep deterministic policy gradient algorithm (MADDPG), for robust policy learning; (2) since the continuous action space leads to computational intractability in our minimax learning objective, we propose Multi-Agent Adversarial Learning (MAAL) to efficiently solve our proposed formulation. We empirically evaluate our M3DDPG algorithm in four mixed cooperative and competitive multi-agent environments and the agents trained by our method significantly outperforms existing baselines.",
"Deep Reinforcement Learning has yielded proficient controllers for complex tasks. However, these controllers have limited memory and rely on being able to perceive the complete game screen at each decision point. To address these shortcomings, this article investigates the effects of adding recurrency to a Deep Q-Network (DQN) by replacing the first post-convolutional fully-connected layer with a recurrent LSTM. The resulting (DRQN), although capable of seeing only a single frame at each timestep, successfully integrates information through time and replicates DQN's performance on standard Atari games and partially observed equivalents featuring flickering game screens. Additionally, when trained with partial observations and evaluated with incrementally more complete observations, DRQN's performance scales as a function of observability. Conversely, when trained with full observations and evaluated with partial observations, DRQN's performance degrades less than DQN's. Thus, given the same length of history, recurrency is a viable alternative to stacking a history of frames in the DQN's input layer and while recurrency confers no systematic advantage when learning to play the game, the recurrent net can better adapt at evaluation time if the quality of observations changes."
]
} |
1907.09597 | 2963065757 | In this paper we explore how actor-critic methods in deep reinforcement learning, in particular Asynchronous Advantage Actor-Critic (A3C), can be extended with agent modeling. Inspired by recent works on representation learning and multiagent deep reinforcement learning, we propose two architectures to perform agent modeling: the first one based on parameter sharing, and the second one based on agent policy features. Both architectures aim to learn other agents' policies as auxiliary tasks, besides the standard actor (policy) and critic (values). We performed experiments in both cooperative and competitive domains. The former is a problem of coordinated multiagent object transportation and the latter is a two-player mini version of the Pommerman game. Our results show that the proposed architectures stabilize learning and outperform the standard A3C architecture when learning a best response in terms of expected rewards. | Deep Cognitive Hierarchies @cite_11 is an algorithm that aims to avoid overfitting in two-player games. It uses deep reinforcement learning to compute best responses to a distribution over policies and empirical game-theoretic analysis to compute new meta-strategy distributions. Theory of Mind Network @cite_16 tackles the problem of meta-learning, i.e., the proposed network should acquire a strong prior model for agents’ behaviour to bootstrap to richer predictions. DeepBPR+ studies the problem of efficient policy detection and reuse when playing against non-stationary agents in Markov games @cite_20 . In contrast, our goal is to estimate the opponent teammate's policy at the same time that the agent is learning its respective (best response) policy; since these two elements are linked to each other our proposals improve the stability of the learning process as well as increase the obtained rewards. | {
"cite_N": [
"@cite_16",
"@cite_20",
"@cite_11"
],
"mid": [
"2963937357",
"2807340089",
"2963642149"
],
"abstract": [
"There has been a resurgence of interest in multiagent reinforcement learning (MARL), due partly to the recent success of deep neural networks. The simplest form of MARL is independent reinforcement learning (InRL), where each agent treats all of its experience as part of its (non stationary) environment. In this paper, we first observe that policies learned using InRL can overfit to the other agents' policies during training, failing to sufficiently generalize during execution. We introduce a new metric, joint-policy correlation, to quantify this effect. We describe a meta-algorithm for general MARL, based on approximate best responses to mixtures of policies generated using deep reinforcement learning, and empirical game theoretic analysis to compute meta-strategies for policy selection. The meta-algorithm generalizes previous algorithms such as InRL, iterated best response, double oracle, and fictitious play. Then, we propose a scalable implementation which reduces the memory requirement using decoupled meta-solvers. Finally, we demonstrate the generality of the resulting policies in three partially observable settings: gridworld coordination problems, emergent language games, and poker.",
"We introduce an approach for deep reinforcement learning (RL) that improves upon the efficiency, generalization capacity, and interpretability of conventional approaches through structured perception and relational reasoning. It uses self-attention to iteratively reason about the relations between entities in a scene and to guide a model-free policy. Our results show that in a novel navigation and planning task called Box-World, our agent finds interpretable solutions that improve upon baselines in terms of sample complexity, ability to generalize to more complex scenes than experienced during training, and overall performance. In the StarCraft II Learning Environment, our agent achieves state-of-the-art performance on six mini-games -- surpassing human grandmaster performance on four. By considering architectural inductive biases, our work opens new directions for overcoming important, but stubborn, challenges in deep RL.",
"Sequential decision making problems, such as structured prediction, robotic control, and game playing, require a combination of planning policies and generalisation of those plans. In this paper, we present Expert Iteration (ExIt), a novel reinforcement learning algorithm which decomposes the problem into separate planning and generalisation tasks. Planning new policies is performed by tree search, while a deep neural network generalises those plans. Subsequently, tree search is improved by using the neural network policy to guide search, increasing the strength of new plans. In contrast, standard deep Reinforcement Learning algorithms rely on a neural network not only to generalise plans, but to discover them too. We show that ExIt outperforms REINFORCE for training a neural network to play the board game Hex, and our final tree search agent, trained tabula rasa, defeats MoHex1.0, the most recent Olympiad Champion player to be publicly released."
]
} |
1907.09624 | 2964198301 | Object classes that surround us have a natural tendency to emerge at varying levels of abstraction. We propose a Bayesian approach to zero-shot learning (ZSL) that introduces the notion of meta-classes and implements a Bayesian hierarchy around these classes to effectively blend data likelihood with local and global priors. Local priors driven by data from seen classes, i.e. classes that are available at training time, become instrumental in recovering unseen classes, i.e. classes that are missing at training time, in a generalized ZSL setting. Hyperparameters of the Bayesian model offer a convenient way to optimize the trade-off between seen and unseen class accuracy in addition to guiding other aspects of model fitting. We conduct experiments on seven benchmark datasets including the large scale ImageNet and show that our model improves the current state of the art in the challenging generalized ZSL setting. | Generative models for zero-shot learning. Although most of the early work focused on discriminative models there are a few studies that use generative models to tackle ZSL @cite_22 @cite_0 . The study in @cite_22 uses Normal distributions to model both image features and semantic vectors and learns a multimodal mapping between two spaces. This mapping is optimized by minimizing a similarity based cross domain loss function. In a similar fashion the study in @cite_0 utilizes a regression model to optimize a mapping between class attributes and parameters of class conditional distributions. A comprehensive review of these techniques and their performance on several benchmark data sets can be found in @cite_6 . | {
"cite_N": [
"@cite_0",
"@cite_22",
"@cite_6"
],
"mid": [
"2516449915",
"2618752949",
"2520613337"
],
"abstract": [
"With the sustaining bloom of multimedia data, Zero-shot Learning (ZSL) techniques have attracted much attention in recent years for its ability to train learning models that can handle “unseen” categories. Existing ZSL algorithms mainly take advantages of attribute-based semantic space and only focus on static image data. Besides, most ZSL studies merely consider the semantic embedded labels and fail to address domain shift problem. In this paper, we purpose a deep two-output model for video ZSL and action recognition tasks by computing both spatial and temporal features from video contents through distinct Convolutional Neural Networks (CNNs) and training a Multi-layer Perceptron (MLP) upon extracted features to map videos to semantic embedding word vectors. Moreover, we introduce a domain adaptation strategy named “ConSSEV” — by combining outputs from two distinct output layers of our MLP to improve the results of zero-shot learning. Our experiments on UCF101 dataset demonstrate the purposed model has more advantages associated with more complex video embedding schemes, and outperforms the state-of-the-art zero-shot learning techniques.",
"Zero-shot learning, which studies the problem of object classification for categories for which we have no training examples, is gaining increasing attention from community. Most existing ZSL methods exploit deterministic transfer learning via an in-between semantic embedding space. In this paper, we try to attack this problem from a generative probabilistic modelling perspective. We assume for any category, the observed representation, e.g. images or texts, is developed from a unique prototype in a latent space, in which the semantic relationship among prototypes is encoded via linear reconstruction. Taking advantage of this assumption, virtual instances of unseen classes can be generated from the corresponding prototype, giving rise to a novel ZSL model which can alleviate the domain shift problem existing in the way of direct transfer learning. Extensive experiments on three benchmark datasets show our proposed model can achieve state-of-the-art results.",
"Zero-Shot Learning (ZSL) promises to scale visual recognition by bypassing the conventional model training requirement of annotated examples for every category. This is achieved by establishing a mapping connecting low-level features and a semantic description of the label space, referred as visual-semantic mapping, on auxiliary data. Re-using the learned mapping to project target videos into an embedding space thus allows novel-classes to be recognised by nearest neighbour inference. However, existing ZSL methods suffer from auxiliary-target domain shift intrinsically induced by assuming the same mapping for the disjoint auxiliary and target classes. This compromises the generalisation accuracy of ZSL recognition on the target data. In this work, we improve the ability of ZSL to generalise across this domain shift in both model- and data-centric ways by formulating a visual-semantic mapping with better generalisation properties and a dynamic data re-weighting method to prioritise auxiliary data that are relevant to the target classes. Specifically: (1) We introduce a multi-task visual-semantic mapping to improve generalisation by constraining the semantic mapping parameters to lie on a low-dimensional manifold, (2) We explore prioritised data augmentation by expanding the pool of auxiliary data with additional instances weighted by relevance to the target domain. The proposed new model is applied to the challenging zero-shot action recognition problem to demonstrate its advantages over existing ZSL models."
]
} |
1907.09624 | 2964198301 | Object classes that surround us have a natural tendency to emerge at varying levels of abstraction. We propose a Bayesian approach to zero-shot learning (ZSL) that introduces the notion of meta-classes and implements a Bayesian hierarchy around these classes to effectively blend data likelihood with local and global priors. Local priors driven by data from seen classes, i.e. classes that are available at training time, become instrumental in recovering unseen classes, i.e. classes that are missing at training time, in a generalized ZSL setting. Hyperparameters of the Bayesian model offer a convenient way to optimize the trade-off between seen and unseen class accuracy in addition to guiding other aspects of model fitting. We conduct experiments on seven benchmark datasets including the large scale ImageNet and show that our model improves the current state of the art in the challenging generalized ZSL setting. | Another close line of work leveraging hierarchical Bayesian model is done in @cite_3 . Their method is similar to ours in the way priors are used to achieve knowledge transfer across classes. However, unlike ours, no semantic information is used when establishing the Bayesian hierarchy in @cite_3 and class discovery is performed in a fully unsupervised fashion. Also, @math and @math play critical roles in modeling expected global and local class dispersions (not modeled in [A1]). | {
"cite_N": [
"@cite_3"
],
"mid": [
"2150385401"
],
"abstract": [
"We develop a hierarchical Bayesian model that learns categories from single training examples. The model transfers acquired knowledge from previously learned categories to a novel category, in the form of a prior over category means and variances. The model discovers how to group categories into meaningful super-categories that express different priors for new classes. Given a single example of a novel category, we can efficiently infer which super-category the novel category belongs to, and thereby estimate not only the new category's mean but also an appropriate similarity metric based on parameters inherited from the super-category. On MNIST and MSR Cambridge image datasets the model learns useful representations of novel categories based on just a single training example, and performs significantly better than simpler hierarchical Bayesian approaches. It can also discover new categories in a completely unsupervised fashion, given just one or a few examples."
]
} |
1907.09658 | 2962765560 | Although skeleton-based action recognition has achieved great success in recent years, most of the existing methods may suffer from a large model size and slow execution speed. To alleviate this issue, we analyze skeleton sequence properties to propose a Double-feature Double-motion Network (DD-Net) for skeleton-based action recognition. By using a lightweight network structure (i.e., 0.15 million parameters), DD-Net can reach a super fast speed, as 3,500 FPS on one GPU, or, 2,000 FPS on one CPU. By employing robust features, DD-Net achieves the state-of-the-art performance on our experimental datasets: SHREC (i.e., hand actions) and JHMDB (i.e., body actions). Our code will be released with this paper later. | Nowadays, with the fast advancement of deep learning, skeleton acquisition is not limited to use motion capture systems @cite_25 and depth cameras @cite_37 . The RGB data, for instance, can be used to infer 2D skeletons @cite_27 @cite_30 or 3D skeletons @cite_38 @cite_43 in real time. Moreover, even WiFi signals can be used to estimate skeleton data @cite_40 @cite_17 . Those achievements have made skeleton-based action recognition available on a huge amount of multimedia resources and therefore have stimulated the model's development. | {
"cite_N": [
"@cite_30",
"@cite_38",
"@cite_37",
"@cite_43",
"@cite_27",
"@cite_40",
"@cite_25",
"@cite_17"
],
"mid": [
"2770191827",
"2797382244",
"2606294640",
"2953181561",
"2060816892",
"2950007497",
"2079846689",
"2776217220"
],
"abstract": [
"The motion analysis of human skeletons is crucial for human action recognition, which is one of the most active topics in computer vision. In this paper, we propose a fully end-to-end action-attending graphic neural network (A2GNN) for skeleton-based action recognition, in which each irregular skeleton is structured as an undirected attribute graph. To extract high-level semantic representation from skeletons, we perform the local spectral graph filtering on the constructed attribute graphs like the standard image convolution operation. Considering not all joints are informative for action analysis, we design an action-attending layer to detect those salient action units by adaptively weighting skelet al joints. Herein, the filtering responses are parameterized into a weighting function irrelevant to the order of input nodes. To further encode continuous motion variations, the deep features learnt from skelet al graphs are gathered along consecutive temporal slices and then fed into a recurrent gated network. Finally, the spectral graph filtering, action-attending, and recurrent temporal encoding are integrated together to jointly train for the sake of robust action recognition as well as the intelligibility of human actions. To evaluate our A2GNN, we conduct extensive experiments on four benchmark skeleton-based action datasets, including the large-scale challenging NTU RGB+D dataset. The experimental results demonstrate that our network achieves the state-of-the-art performances.",
"Action recognition with 3D skeleton sequences became popular due to its speed and robustness. The recently proposed convolutional neural networks (CNNs)-based methods show a good performance in learning spatio–temporal representations for skeleton sequences. Despite the good recognition accuracy achieved by previous CNN-based methods, there existed two problems that potentially limit the performance. First, previous skeleton representations were generated by chaining joints with a fixed order. The corresponding semantic meaning was unclear and the structural information among the joints was lost. Second, previous models did not have an ability to focus on informative joints. The attention mechanism was important for skeleton-based action recognition because different joints contributed unequally toward the correct recognition. To solve these two problems, we proposed a novel CNN-based method for skeleton-based action recognition. We first redesigned the skeleton representations with a depth-first tree traversal order, which enhanced the semantic meaning of skeleton images and better preserved the associated structural information. We then proposed the general two-branch attention architecture that automatically focused on spatio–temporal key stages and filtered out unreliable joint predictions. Based on the proposed general architecture, we designed a global long-sequence attention network with refined branch structures. Furthermore, in order to adjust the kernel’s spatio–temporal aspect ratios and better capture long-term dependencies, we proposed a sub-sequence attention network (SSAN) that took sub-image sequences as inputs. We showed that the two-branch attention architecture could be combined with the SSAN to further improve the performance. Our experiment results on the NTU RGB+D data set and the SBU kinetic interaction data set outperformed the state of the art. The model was further validated on noisy estimated poses from the subsets of the UCF101 data set and the kinetics data set.",
"Recently, skeleton based action recognition gains more popularity due to cost-effective depth sensors coupled with real-time skeleton estimation algorithms. Traditional approaches based on handcrafted features are limited to represent the complexity of motion patterns. Recent methods that use Recurrent Neural Networks (RNN) to handle raw skeletons only focus on the contextual dependency in the temporal domain and neglect the spatial configurations of articulated skeletons. In this paper, we propose a novel two-stream RNN architecture to model both temporal dynamics and spatial configurations for skeleton based action recognition. We explore two different structures for the temporal stream: stacked RNN and hierarchical RNN. Hierarchical RNN is designed according to human body kinematics. We also propose two effective methods to model the spatial structure by converting the spatial graph into a sequence of joints. To improve generalization of our model, we further exploit 3D transformation based data augmentation techniques including rotation and scaling transformation to transform the 3D coordinates of skeletons during training. Experiments on 3D action recognition benchmark datasets show that our method brings a considerable improvement for a variety of actions, i.e., generic actions, interaction activities and gestures.",
"Recently, skeleton based action recognition gains more popularity due to cost-effective depth sensors coupled with real-time skeleton estimation algorithms. Traditional approaches based on handcrafted features are limited to represent the complexity of motion patterns. Recent methods that use Recurrent Neural Networks (RNN) to handle raw skeletons only focus on the contextual dependency in the temporal domain and neglect the spatial configurations of articulated skeletons. In this paper, we propose a novel two-stream RNN architecture to model both temporal dynamics and spatial configurations for skeleton based action recognition. We explore two different structures for the temporal stream: stacked RNN and hierarchical RNN. Hierarchical RNN is designed according to human body kinematics. We also propose two effective methods to model the spatial structure by converting the spatial graph into a sequence of joints. To improve generalization of our model, we further exploit 3D transformation based data augmentation techniques including rotation and scaling transformation to transform the 3D coordinates of skeletons during training. Experiments on 3D action recognition benchmark datasets show that our method brings a considerable improvement for a variety of actions, i.e., generic actions, interaction activities and gestures.",
"Recent development in affordable depth sensors opens new possibilities in action recognition problem. Depth information improves skeleton detection, therefore many authors focused on analyzing pose for action recognition. But still skeleton detection is not robust and fail in more challenging scenarios, where sensor is placed outside of optimal working range and serious occlusions occur. In this paper we investigate state-of-the-art methods designed for RGB videos, which have proved their performance. Then we extend current state-of-the-art algorithms to benefit from depth information without need of skeleton detection. In this paper we propose two novel video descriptors. First combines motion and 3D information. Second improves performance on actions with low movement rate. We validate our approach on challenging MSR Daily Activty 3D dataset.",
"Motion capture data digitally represent human movements by sequences of 3D skeleton configurations. Such spatio-temporal data, often recorded in the stream-based nature, need to be efficiently processed to detect high-interest actions, for example, in human-computer interaction to understand hand gestures in real time. Alternatively, automatically annotated parts of a continuous stream can be persistently stored to become searchable, and thus reusable for future retrieval or pattern mining. In this paper, we focus on multi-label detection of user-specified actions in unsegmented sequences as well as continuous streams. In particular, we utilize the current advances in recurrent neural networks and adopt a unidirectional LSTM model to effectively encode the skeleton frames within the hidden network states. The model learns what subsequences of encoded frames belong to the specified action classes within the training phase. The learned representations of classes are then employed within the annotation phase to infer the probability that an incoming skeleton frame belongs to a given action class. The computed probabilities are finally compared against a learned threshold to automatically determine the beginnings and endings of actions. To further enhance the annotation accuracy, we utilize a bidirectional LSTM model to estimate class probabilities by considering not only the past frames but also the future ones. We extensively evaluate both the models on the three use cases of real-time stream annotation, offline annotation of long sequences, and early action detection and prediction. The experiments demonstrate that our models outperform the state of the art in effectiveness and are at least one order of magnitude more efficient, being able to annotate 10 k frames per second.",
"We present a novel method for accurate marker-less capture of articulated skeleton motion of several subjects in general scenes, indoors and outdoors, even from input filmed with as few as two cameras. Our approach unites a discriminative image-based joint detection method with a model-based generative motion tracking algorithm through a combined pose optimization energy. The discriminative part-based pose detection method, implemented using Convolutional Networks (ConvNet), estimates unary potentials for each joint of a kinematic skeleton model. These unary potentials are used to probabilistically extract pose constraints for tracking by using weighted sampling from a pose posterior guided by the model. In the final energy, these constraints are combined with an appearance-based model-to-image similarity term. Poses can be computed very efficiently using iterative local optimization, as ConvNet detection is fast, and our formulation yields a combined pose estimation energy with analytic derivatives. In combination, this enables to track full articulated joint angles at state-of-the-art accuracy and temporal stability with a very low number of cameras.",
"Depth sensors open up possibilities of dealing with the human action recognition problem by providing 3D human skeleton data and depth images of the scene. Analysis of human actions based on 3D skeleton data has become popular recently, due to its robustness and view-invariant representation. However, the skeleton alone is insufficient to distinguish actions which involve human-object interactions. In this paper, we propose a deep model which efficiently models human-object interactions and intra-class variations under viewpoint changes. First, a human body-part model is introduced to transfer the depth appearances of body-parts to a shared view-invariant space. Second, an end-to-end learning framework is proposed which is able to effectively combine the view-invariant body-part representation from skelet al and depth images, and learn the relations between the human body-parts and the environmental objects, the interactions between different human body-parts, and the temporal structure of human actions. We have evaluated the performance of our proposed model against 15 existing techniques on two large benchmark human action recognition datasets including NTU RGB+D and UWA3DII. The Experimental results show that our technique provides a significant improvement over state-of-the-art methods."
]
} |
1907.09658 | 2962765560 | Although skeleton-based action recognition has achieved great success in recent years, most of the existing methods may suffer from a large model size and slow execution speed. To alleviate this issue, we analyze skeleton sequence properties to propose a Double-feature Double-motion Network (DD-Net) for skeleton-based action recognition. By using a lightweight network structure (i.e., 0.15 million parameters), DD-Net can reach a super fast speed, as 3,500 FPS on one GPU, or, 2,000 FPS on one CPU. By employing robust features, DD-Net achieves the state-of-the-art performance on our experimental datasets: SHREC (i.e., hand actions) and JHMDB (i.e., body actions). Our code will be released with this paper later. | In general, in order to achieve a better performance for skeleton-based action recognition, previous studies attempt to work on two aspects: introduce new features for skeleton sequences @cite_28 @cite_7 @cite_33 @cite_18 @cite_36 @cite_14 @cite_35 , and, propose novel neural network architectures @cite_2 @cite_34 @cite_44 @cite_29 @cite_46 @cite_39 @cite_1 . | {
"cite_N": [
"@cite_35",
"@cite_18",
"@cite_14",
"@cite_33",
"@cite_7",
"@cite_28",
"@cite_36",
"@cite_29",
"@cite_1",
"@cite_39",
"@cite_44",
"@cite_2",
"@cite_46",
"@cite_34"
],
"mid": [
"2797382244",
"2803158089",
"2606294640",
"2953181561",
"2604321021",
"2415469094",
"2307035320",
"2952587893",
"2341313195",
"2603861860",
"2716916105",
"2770191827",
"2031334527",
"2563154851"
],
"abstract": [
"Action recognition with 3D skeleton sequences became popular due to its speed and robustness. The recently proposed convolutional neural networks (CNNs)-based methods show a good performance in learning spatio–temporal representations for skeleton sequences. Despite the good recognition accuracy achieved by previous CNN-based methods, there existed two problems that potentially limit the performance. First, previous skeleton representations were generated by chaining joints with a fixed order. The corresponding semantic meaning was unclear and the structural information among the joints was lost. Second, previous models did not have an ability to focus on informative joints. The attention mechanism was important for skeleton-based action recognition because different joints contributed unequally toward the correct recognition. To solve these two problems, we proposed a novel CNN-based method for skeleton-based action recognition. We first redesigned the skeleton representations with a depth-first tree traversal order, which enhanced the semantic meaning of skeleton images and better preserved the associated structural information. We then proposed the general two-branch attention architecture that automatically focused on spatio–temporal key stages and filtered out unreliable joint predictions. Based on the proposed general architecture, we designed a global long-sequence attention network with refined branch structures. Furthermore, in order to adjust the kernel’s spatio–temporal aspect ratios and better capture long-term dependencies, we proposed a sub-sequence attention network (SSAN) that took sub-image sequences as inputs. We showed that the two-branch attention architecture could be combined with the SSAN to further improve the performance. Our experiment results on the NTU RGB+D data set and the SBU kinetic interaction data set outperformed the state of the art. The model was further validated on noisy estimated poses from the subsets of the UCF101 data set and the kinetics data set.",
"Recently, skeleton-based action recognition becomes popular owing to the development of cost-effective depth sensors and fast pose estimation algorithms. Traditional methods based on pose descriptors often fail on large-scale datasets due to the limited representation of engineered features. Recent recurrent neural networks (RNN) based approaches mostly focus on the temporal evolution of body joints and neglect the geometric relations. In this paper, we aim to leverage the geometric relations among joints for action recognition. We introduce three primitive geometries: joints, edges, and surfaces. Accordingly, a generic end-to-end RNN based network is designed to accommodate the three inputs. For action recognition, a novel viewpoint transformation layer and temporal dropout layers are utilized in the RNN based network to learn robust representations. And for action detection, we first perform frame-wise action classification, then exploit a novel multi-scale sliding window algorithm. Experiments on the large-scale 3D action recognition benchmark datasets show that joints, edges, and surfaces are effective and complementary for different actions. Our approaches dramatically outperform the existing state-of-the-art methods for both tasks of action recognition and action detection.",
"Recently, skeleton based action recognition gains more popularity due to cost-effective depth sensors coupled with real-time skeleton estimation algorithms. Traditional approaches based on handcrafted features are limited to represent the complexity of motion patterns. Recent methods that use Recurrent Neural Networks (RNN) to handle raw skeletons only focus on the contextual dependency in the temporal domain and neglect the spatial configurations of articulated skeletons. In this paper, we propose a novel two-stream RNN architecture to model both temporal dynamics and spatial configurations for skeleton based action recognition. We explore two different structures for the temporal stream: stacked RNN and hierarchical RNN. Hierarchical RNN is designed according to human body kinematics. We also propose two effective methods to model the spatial structure by converting the spatial graph into a sequence of joints. To improve generalization of our model, we further exploit 3D transformation based data augmentation techniques including rotation and scaling transformation to transform the 3D coordinates of skeletons during training. Experiments on 3D action recognition benchmark datasets show that our method brings a considerable improvement for a variety of actions, i.e., generic actions, interaction activities and gestures.",
"Recently, skeleton based action recognition gains more popularity due to cost-effective depth sensors coupled with real-time skeleton estimation algorithms. Traditional approaches based on handcrafted features are limited to represent the complexity of motion patterns. Recent methods that use Recurrent Neural Networks (RNN) to handle raw skeletons only focus on the contextual dependency in the temporal domain and neglect the spatial configurations of articulated skeletons. In this paper, we propose a novel two-stream RNN architecture to model both temporal dynamics and spatial configurations for skeleton based action recognition. We explore two different structures for the temporal stream: stacked RNN and hierarchical RNN. Hierarchical RNN is designed according to human body kinematics. We also propose two effective methods to model the spatial structure by converting the spatial graph into a sequence of joints. To improve generalization of our model, we further exploit 3D transformation based data augmentation techniques including rotation and scaling transformation to transform the 3D coordinates of skeletons during training. Experiments on 3D action recognition benchmark datasets show that our method brings a considerable improvement for a variety of actions, i.e., generic actions, interaction activities and gestures.",
"This paper presents a new method for 3D action recognition with skeleton sequences (i.e., 3D trajectories of human skeleton joints). The proposed method first transforms each skeleton sequence into three clips each consisting of several frames for spatial temporal feature learning using deep neural networks. Each clip is generated from one channel of the cylindrical coordinates of the skeleton sequence. Each frame of the generated clips represents the temporal information of the entire skeleton sequence, and incorporates one particular spatial relationship between the joints. The entire clips include multiple frames with different spatial relationships, which provide useful spatial structural information of the human skeleton. We propose to use deep convolutional neural networks to learn long-term temporal information of the skeleton sequence from the frames of the generated clips, and then use a Multi-Task Learning Network (MTLN) to jointly process all frames of the clips in parallel to incorporate spatial structural information for action recognition. Experimental results clearly show the effectiveness of the proposed new representation and feature learning method for 3D action recognition.",
"Temporal dynamics of postures over time is crucial for sequence-based action recognition. Human actions can be represented by the corresponding motions of articulated skeleton. Most of the existing approaches for skeleton based action recognition model the spatial-temporal evolution of actions based on hand-crafted features. As a kind of hierarchically adaptive filter banks, Convolutional Neural Network (CNN) performs well in representation learning. In this paper, we propose an end-to-end hierarchical architecture for skeleton based action recognition with CNN. Firstly, we represent a skeleton sequence as a matrix by concatenating the joint coordinates in each instant and arranging those vector representations in a chronological order. Then the matrix is quantified into an image and normalized to handle the variable-length problem. The final image is fed into a CNN model for feature extraction and recognition. For the specific structure of such images, the simple max-pooling plays an important role on spatial feature selection as well as temporal frequency adjustment, which can obtain more discriminative joint information for different actions and meanwhile address the variable-frequency problem. Experimental results demonstrate that our method achieves the state-of-art performance with high computational efficiency, especially surpassing the existing result by more than 15 percentage on the challenging ChaLearn gesture recognition dataset.",
"Skeleton based action recognition distinguishes human actions using the trajectories of skeleton joints, which provide a very good representation for describing actions. Considering that recurrent neural networks (RNNs) with Long Short-Term Memory (LSTM) can learn feature representations and model long-term temporal dependencies automatically, we propose an end-to-end fully connected deep LSTM network for skeleton based action recognition. Inspired by the observation that the co-occurrences of the joints intrinsically characterize human actions, we take the skeleton as the input at each time slot and introduce a novel regularization scheme to learn the co-occurrence features of skeleton joints. To train the deep LSTM network effectively, we propose a new dropout algorithm which simultaneously operates on the gates, cells, and output responses of the LSTM neurons. Experimental results on three human action recognition datasets consistently demonstrate the effectiveness of the proposed model.",
"Skeleton based action recognition distinguishes human actions using the trajectories of skeleton joints, which provide a very good representation for describing actions. Considering that recurrent neural networks (RNNs) with Long Short-Term Memory (LSTM) can learn feature representations and model long-term temporal dependencies automatically, we propose an end-to-end fully connected deep LSTM network for skeleton based action recognition. Inspired by the observation that the co-occurrences of the joints intrinsically characterize human actions, we take the skeleton as the input at each time slot and introduce a novel regularization scheme to learn the co-occurrence features of skeleton joints. To train the deep LSTM network effectively, we propose a new dropout algorithm which simultaneously operates on the gates, cells, and output responses of the LSTM neurons. Experimental results on three human action recognition datasets consistently demonstrate the effectiveness of the proposed model.",
"Human action recognition from well-segmented 3D skeleton data has been intensively studied and has been attracting an increasing attention. Online action detection goes one step further and is more challenging, which identifies the action type and localizes the action positions on the fly from the untrimmed stream data. In this paper, we study the problem of online action detection from streaming skeleton data. We propose a multi-task end-to-end Joint Classification-Regression Recurrent Neural Network to better explore the action type and temporal localization information. By employing a joint classification and regression optimization objective, this network is capable of automatically localizing the start and end points of actions more accurately. Specifically, by leveraging the merits of the deep Long Short-Term Memory (LSTM) subnetwork, the proposed model automatically captures the complex long-range temporal dynamics, which naturally avoids the typical sliding window design and thus ensures high computational efficiency. Furthermore, the subtask of regression optimization provides the ability to forecast the action prior to its occurrence. To evaluate our proposed model, we build a large streaming video dataset with annotations. Experimental results on our dataset and the public G3D dataset both demonstrate very promising performance of our scheme.",
"Skeleton-based human action recognition has recently attracted increasing attention due to the popularity of 3D skeleton data. One main challenge lies in the large view variations in captured human actions. We propose a novel view adaptation scheme to automatically regulate observation viewpoints during the occurrence of an action. Rather than re-positioning the skeletons based on a human defined prior criterion, we design a view adaptive recurrent neural network (RNN) with LSTM architecture, which enables the network itself to adapt to the most suitable observation viewpoints from end to end. Extensive experiment analyses show that the proposed view adaptive RNN model strives to (1) transform the skeletons of various views to much more consistent viewpoints and (2) maintain the continuity of the action rather than transforming every frame to the same position with the same body orientation. Our model achieves significant improvement over the state-of-the-art approaches on three benchmark datasets.",
"Skeleton-based human action recognition has attracted a lot of research attention during the past few years. Recent works attempted to utilize recurrent neural networks to model the temporal dependencies between the 3D positional configurations of human body joints for better analysis of human activities in the skelet al data. The proposed work extends this idea to spatial domain as well as temporal domain to better analyze the hidden sources of action-related information within the human skeleton sequences in both of these domains simultaneously. Based on the pictorial structure of Kinect's skelet al data, an effective tree-structure based traversal framework is also proposed. In order to deal with the noise in the skelet al data, a new gating mechanism within LSTM module is introduced, with which the network can learn the reliability of the sequential data and accordingly adjust the effect of the input data on the updating procedure of the long-term context representation stored in the unit's memory cell. Moreover, we introduce a novel multi-modal feature fusion strategy within the LSTM unit in this paper. The comprehensive experimental results on seven challenging benchmark datasets for human action recognition demonstrate the effectiveness of the proposed method.",
"The motion analysis of human skeletons is crucial for human action recognition, which is one of the most active topics in computer vision. In this paper, we propose a fully end-to-end action-attending graphic neural network (A2GNN) for skeleton-based action recognition, in which each irregular skeleton is structured as an undirected attribute graph. To extract high-level semantic representation from skeletons, we perform the local spectral graph filtering on the constructed attribute graphs like the standard image convolution operation. Considering not all joints are informative for action analysis, we design an action-attending layer to detect those salient action units by adaptively weighting skelet al joints. Herein, the filtering responses are parameterized into a weighting function irrelevant to the order of input nodes. To further encode continuous motion variations, the deep features learnt from skelet al graphs are gathered along consecutive temporal slices and then fed into a recurrent gated network. Finally, the spectral graph filtering, action-attending, and recurrent temporal encoding are integrated together to jointly train for the sake of robust action recognition as well as the intelligibility of human actions. To evaluate our A2GNN, we conduct extensive experiments on four benchmark skeleton-based action datasets, including the large-scale challenging NTU RGB+D dataset. The experimental results demonstrate that our network achieves the state-of-the-art performances.",
"HighlightsEffective method to recognize human actions using 3D skeleton joints.New action feature descriptor, EigenJoints, for action recognition.Accumulated Motion Energy (AME) method to perform informative frames selection.Our proposed approach significantly outperforms the state-of-the-art methods on three public datasets. In this paper, we propose an effective method to recognize human actions using 3D skeleton joints recovered from 3D depth data of RGBD cameras. We design a new action feature descriptor for action recognition based on differences of skeleton joints, i.e., EigenJoints which combine action information including static posture, motion property, and overall dynamics. Accumulated Motion Energy (AME) is then proposed to perform informative frame selection, which is able to remove noisy frames and reduce computational cost. We employ non-parametric Naive-Bayes-Nearest-Neighbor (NBNN) to classify multiple actions. The experimental results on several challenging datasets demonstrate that our approach outperforms the state-of-the-art methods. In addition, we investigate how many frames are necessary for our method to perform classification in the scenario of online action recognition. We observe that the first 30-40 frames are sufficient to achieve comparable results to that using the entire video sequences on the MSR Action3D dataset.",
"In recent years, skeleton-based action recognition has become a popular 3D classification problem. State-of-the-art methods typically first represent each motion sequence as a high-dimensional trajectory on a Lie group with an additional dynamic time warping, and then shallowly learn favorable Lie group features. In this paper we incorporate the Lie group structure into a deep network architecture to learn more appropriate Lie group features for 3D action recognition. Within the network structure, we design rotation mapping layers to transform the input Lie group features into desirable ones, which are aligned better in the temporal domain. To reduce the high feature dimensionality, the architecture is equipped with rotation pooling layers for the elements on the Lie group. Furthermore, we propose a logarithm mapping layer to map the resulting manifold data into a tangent space that facilitates the application of regular output layers for the final classification. Evaluations of the proposed network for standard 3D human action recognition datasets clearly demonstrate its superiority over existing shallow Lie group feature learning methods as well as most conventional deep learning methods."
]
} |
1907.09658 | 2962765560 | Although skeleton-based action recognition has achieved great success in recent years, most of the existing methods may suffer from a large model size and slow execution speed. To alleviate this issue, we analyze skeleton sequence properties to propose a Double-feature Double-motion Network (DD-Net) for skeleton-based action recognition. By using a lightweight network structure (i.e., 0.15 million parameters), DD-Net can reach a super fast speed, as 3,500 FPS on one GPU, or, 2,000 FPS on one CPU. By employing robust features, DD-Net achieves the state-of-the-art performance on our experimental datasets: SHREC (i.e., hand actions) and JHMDB (i.e., body actions). Our code will be released with this paper later. | A good skeleton-sequence representation should contain global motion information and be location-viewpoint invariant. However, it is challenging to satisfy both requirements in one feature. The studies @cite_7 @cite_18 @cite_42 @cite_35 focused on global motions without considering the location-viewpoint variation in their features. Other studies @cite_28 @cite_33 @cite_36 , on the contrary, introduced location-viewpoint invariant features without considering global motions. Our work bridges their gaps by seamlessly integrating a location-viewpoint invariant feature and a two-scale global motion feature together. | {
"cite_N": [
"@cite_35",
"@cite_18",
"@cite_33",
"@cite_7",
"@cite_28",
"@cite_36",
"@cite_42"
],
"mid": [
"2797382244",
"2593146028",
"2963369114",
"2021150171",
"2146700972",
"2769356789",
"2603861860"
],
"abstract": [
"Action recognition with 3D skeleton sequences became popular due to its speed and robustness. The recently proposed convolutional neural networks (CNNs)-based methods show a good performance in learning spatio–temporal representations for skeleton sequences. Despite the good recognition accuracy achieved by previous CNN-based methods, there existed two problems that potentially limit the performance. First, previous skeleton representations were generated by chaining joints with a fixed order. The corresponding semantic meaning was unclear and the structural information among the joints was lost. Second, previous models did not have an ability to focus on informative joints. The attention mechanism was important for skeleton-based action recognition because different joints contributed unequally toward the correct recognition. To solve these two problems, we proposed a novel CNN-based method for skeleton-based action recognition. We first redesigned the skeleton representations with a depth-first tree traversal order, which enhanced the semantic meaning of skeleton images and better preserved the associated structural information. We then proposed the general two-branch attention architecture that automatically focused on spatio–temporal key stages and filtered out unreliable joint predictions. Based on the proposed general architecture, we designed a global long-sequence attention network with refined branch structures. Furthermore, in order to adjust the kernel’s spatio–temporal aspect ratios and better capture long-term dependencies, we proposed a sub-sequence attention network (SSAN) that took sub-image sequences as inputs. We showed that the two-branch attention architecture could be combined with the SSAN to further improve the performance. Our experiment results on the NTU RGB+D data set and the SBU kinetic interaction data set outperformed the state of the art. The model was further validated on noisy estimated poses from the subsets of the UCF101 data set and the kinetics data set.",
"Sequence-based view invariant transform can effectively cope with view variations.Enhanced skeleton visualization method encodes spatio-temporal skeletons as visual and motion enhanced color images in a compact yet distinctive manner.Multi-stream convolutional neural networks fusion model is able to explore complementary properties among different types of enhanced color images.Our method consistently achieves the highest accuracies on four datasets, including the largest and most challenging NTU RGB+D dataset for skeleton-based action recognition. Human action recognition based on skeletons has wide applications in humancomputer interaction and intelligent surveillance. However, view variations and noisy data bring challenges to this task. Whats more, it remains a problem to effectively represent spatio-temporal skeleton sequences. To solve these problems in one goal, this work presents an enhanced skeleton visualization method for view invariant human action recognition. Our method consists of three stages. First, a sequence-based view invariant transform is developed to eliminate the effect of view variations on spatio-temporal locations of skeleton joints. Second, the transformed skeletons are visualized as a series of color images, which implicitly encode the spatio-temporal information of skeleton joints. Furthermore, visual and motion enhancement methods are applied on color images to enhance their local patterns. Third, a convolutional neural networks-based model is adopted to extract robust and discriminative features from color images. The final action class scores are generated by decision level fusion of deep features. Extensive experiments on four challenging datasets consistently demonstrate the superiority of our method.",
"Skeleton-based human action recognition has recently drawn increasing attentions with the availability of large-scale skeleton datasets. The most crucial factors for this task lie in two aspects: the intra-frame representation for joint co-occurrences and the inter-frame representation for skeletons' temporal evolutions. In this paper we propose an end-to-end convolutional co-occurrence feature learning framework. The co-occurrence features are learned with a hierarchical methodology, in which different levels of contextual information are aggregated gradually. Firstly point-level information of each joint is encoded independently. Then they are assembled into semantic representation in both spatial and temporal domains. Specifically, we introduce a global spatial aggregation scheme, which is able to learn superior joint co-occurrence features over local aggregation. Besides, raw skeleton coordinates as well as their temporal difference are integrated with a two-stream paradigm. Experiments show that our approach consistently outperforms other state-of-the-arts on action recognition and detection benchmarks like NTU RGB+D, SBU Kinect Interaction and PKU-MMD.",
"Recent advances on human motion analysis have made the extraction of human skeleton structure feasible, even from single depth images. This structure has been proven quite informative for discriminating actions in a recognition scenario. In this context, we propose a local skeleton descriptor that encodes the relative position of joint quadruples. Such a coding implies a similarity normalisation transform that leads to a compact (6D) view-invariant skelet al feature, referred to as skelet al quad. Further, the use of a Fisher kernel representation is suggested to describe the skelet al quads contained in a (sub)action. A Gaussian mixture model is learnt from training data, so that the generation of any set of quads is encoded by its Fisher vector. Finally, a multi-level representation of Fisher vectors leads to an action description that roughly carries the order of sub-action within each action sequence. Efficient classification is here achieved by linear SVMs. The proposed action representation is tested on widely used datasets, MSRAction3D and HDM05. The experimental evaluation shows that the proposed method outperforms state-of-the-art algorithms that rely only on joints, while it competes with methods that combine joints with extra cues.",
"We introduce the notion of consensus skeletons for non-rigid space-time registration of a deforming shape. Instead of basing the registration on point features, which are local and sensitive to noise, we adopt the curve skeleton of the shape as a global and descriptive feature for the task. Our method uses no template and only assumes that the skelet al structure of the captured shape remains largely consistent over time. Such an assumption is generally weaker than those relying on large overlap of point features between successive frames, allowing for more sparse acquisition across time. Building our registration framework on top of the low-dimensional skeletontime structure avoids heavy processing of dense point or volumetric data, while skeleton consensusization provides robust handling of incompatibilities between per-frame skeletons. To register point clouds from all frames, we deform them by their skeletons, mirroring the skeleton registration process, to jump-start a non-rigid ICP. We present results for non-rigid space-time registration under sparse and noisy spatio-temporal sampling, including cases where data was captured from only a single view.",
"Human skeleton joints are popular for action analysis since they can be easily extracted from videos to discard background noises. However, current skeleton representations do not fully benefit from machine learning with CNNs. We propose \"Skepxels\" a spatio-temporal representation for skeleton sequences to fully exploit the \"local\" correlations between joints using the 2D convolution kernels of CNN. We transform skeleton videos into images of flexible dimensions using Skepxels and develop a CNN-based framework for effective human action recognition using the resulting images. Skepxels encode rich spatio-temporal information about the skeleton joints in the frames by maximizing a unique distance metric, defined collaboratively over the distinct joint arrangements used in the skelet al image. Moreover, they are flexible in encoding compound semantic notions such as location and speed of the joints. The proposed action recognition exploits the representation in a hierarchical manner by first capturing the micro-temporal relations between the skeleton joints with the Skepxels and then exploiting their macro-temporal relations by computing the Fourier Temporal Pyramids over the CNN features of the skelet al images. We extend the Inception-ResNet CNN architecture with the proposed method and improve the state-of-the-art accuracy by 4.4 on the large scale NTU human activity dataset. On the medium-sized N-UCLA and UTH-MHAD datasets, our method outperforms the existing results by 5.7 and 9.3 respectively.",
"Skeleton-based human action recognition has recently attracted increasing attention due to the popularity of 3D skeleton data. One main challenge lies in the large view variations in captured human actions. We propose a novel view adaptation scheme to automatically regulate observation viewpoints during the occurrence of an action. Rather than re-positioning the skeletons based on a human defined prior criterion, we design a view adaptive recurrent neural network (RNN) with LSTM architecture, which enables the network itself to adapt to the most suitable observation viewpoints from end to end. Extensive experiment analyses show that the proposed view adaptive RNN model strives to (1) transform the skeletons of various views to much more consistent viewpoints and (2) maintain the continuity of the action rather than transforming every frame to the same position with the same body orientation. Our model achieves significant improvement over the state-of-the-art approaches on three benchmark datasets."
]
} |
1907.09658 | 2962765560 | Although skeleton-based action recognition has achieved great success in recent years, most of the existing methods may suffer from a large model size and slow execution speed. To alleviate this issue, we analyze skeleton sequence properties to propose a Double-feature Double-motion Network (DD-Net) for skeleton-based action recognition. By using a lightweight network structure (i.e., 0.15 million parameters), DD-Net can reach a super fast speed, as 3,500 FPS on one GPU, or, 2,000 FPS on one CPU. By employing robust features, DD-Net achieves the state-of-the-art performance on our experimental datasets: SHREC (i.e., hand actions) and JHMDB (i.e., body actions). Our code will be released with this paper later. | Although Recurrent Neural Networks (RNNs) are commonly used in skeleton-based action recognition @cite_26 @cite_24 @cite_20 @cite_23 @cite_21 @cite_45 , we argue that it is relatively slow and difficult for parallel computing, compared with methods @cite_2 @cite_41 @cite_35 that use Convolutional Neural Networks (CNNs). Since we take the model speed as one of our priorities, we utilize 1D CNNs to construct the backbone network of DD-Net. | {
"cite_N": [
"@cite_35",
"@cite_26",
"@cite_41",
"@cite_21",
"@cite_24",
"@cite_45",
"@cite_23",
"@cite_2",
"@cite_20"
],
"mid": [
"2803158089",
"2787900668",
"2799034895",
"1922658220",
"2779380177",
"2606294640",
"2953181561",
"1934184906",
"1950788856"
],
"abstract": [
"Recently, skeleton-based action recognition becomes popular owing to the development of cost-effective depth sensors and fast pose estimation algorithms. Traditional methods based on pose descriptors often fail on large-scale datasets due to the limited representation of engineered features. Recent recurrent neural networks (RNN) based approaches mostly focus on the temporal evolution of body joints and neglect the geometric relations. In this paper, we aim to leverage the geometric relations among joints for action recognition. We introduce three primitive geometries: joints, edges, and surfaces. Accordingly, a generic end-to-end RNN based network is designed to accommodate the three inputs. For action recognition, a novel viewpoint transformation layer and temporal dropout layers are utilized in the RNN based network to learn robust representations. And for action detection, we first perform frame-wise action classification, then exploit a novel multi-scale sliding window algorithm. Experiments on the large-scale 3D action recognition benchmark datasets show that joints, edges, and surfaces are effective and complementary for different actions. Our approaches dramatically outperform the existing state-of-the-art methods for both tasks of action recognition and action detection.",
"Deep convolutional neural networks (CNNs) have made impressive progress in many video recognition tasks such as video pose estimation and video object detection. However, CNN inference on video is computationally expensive due to processing dense frames individually. In this work, we propose a framework called Recurrent Residual Module (RRM) to accelerate the CNN inference for video recognition tasks. This framework has a novel design of using the similarity of the intermediate feature maps of two consecutive frames, to largely reduce the redundant computation. One unique property of the proposed method compared to previous work is that feature maps of each frame are precisely computed. The experiments show that, while maintaining the similar recognition performance, our RRM yields averagely 2x acceleration on the commonly used CNNs such as AlexNet, ResNet, deep compression model (thus 8-12x faster than the original dense models using the efficient inference engine), and impressively 9x acceleration on some binary networks such as XNOR-Nets (thus 500x faster than the original model). We further verify the effectiveness of the RRM on speeding up CNNs for video pose estimation and video object detection.",
"Recurrent neural networks (RNNs) have emerged as a powerful model for a broad range of machine learning problems that involve sequential data. While an abundance of work exists to understand and improve RNNs in the context of language and audio signals such as language modeling and speech recognition, relatively little attention has been paid to analyze or modify RNNs for visual sequences, which by nature have distinct properties. In this paper, we aim to bridge this gap and present the first large-scale exploration of RNNs for visual sequence learning. In particular, with the intention of leveraging the strong generalization capacity of pre-trained convolutional neural networks (CNNs), we propose a novel and effective approach, PreRNN, to make pre-trained CNNs recurrent by transforming convolutional layers or fully connected layers into recurrent layers. We conduct extensive evaluations on three representative visual sequence learning tasks: sequential face alignment, dynamic hand gesture recognition, and action recognition. Our experiments reveal that PreRNN consistently outperforms the traditional RNNs and achieves state-of-the-art results on the three applications, suggesting that PreRNN is more suitable for visual sequence learning.",
"In existing convolutional neural networks (CNNs), both convolution and pooling are locally performed for image regions separately, no contextual dependencies between different image regions have been taken into consideration. Such dependencies represent useful spatial structure information in images. Whereas recurrent neural networks (RNNs) are designed for learning contextual dependencies among sequential data by using the recurrent (feedback) connections. In this work, we propose the convolutional recurrent neural network (C-RNN), which learns the spatial dependencies between image regions to enhance the discriminative power of image representation. The C-RNN is trained in an end-to-end manner from raw pixel images. CNN layers are firstly processed to generate middle level features. RNN layer is then learned to encode spatial dependencies. The C-RNN can learn better image representation, especially for images with obvious spatial contextual dependencies. Our method achieves competitive performance on ILSVRC 2012, SUN 397, and MIT indoor.",
"Recent studies demonstrate the effectiveness of Recurrent Neural Networks (RNNs) for action recognition in videos. However, previous works mainly utilize video-level category as supervision to train RNNs, which may prohibit RNNs to learn complex motion structures along time. In this paper, we propose a recurrent pose-attention network (RPAN) to address this challenge, where we introduce a novel pose-attention mechanism to adaptively learn pose-related features at every time-step action prediction of RNNs. More specifically, we make three main contributions in this paper. Firstly, unlike previous works on pose-related action recognition, our RPAN is an end-toend recurrent network which can exploit important spatialtemporal evolutions of human pose to assist action recognition in a unified framework. Secondly, instead of learning individual human-joint features separately, our poseattention mechanism learns robust human-part features by sharing attention parameters partially on the semanticallyrelated human joints. These human-part features are then fed into the human-part pooling layer to construct a highlydiscriminative pose-related representation for temporal action modeling. Thirdly, one important byproduct of our RPAN is pose estimation in videos, which can be used for coarse pose annotation in action videos. We evaluate the proposed RPAN quantitatively and qualitatively on two popular benchmarks, i.e., Sub-JHMDB and PennAction. Experimental results show that RPAN outperforms the recent state-of-the-art methods on these challenging datasets.",
"Recently, skeleton based action recognition gains more popularity due to cost-effective depth sensors coupled with real-time skeleton estimation algorithms. Traditional approaches based on handcrafted features are limited to represent the complexity of motion patterns. Recent methods that use Recurrent Neural Networks (RNN) to handle raw skeletons only focus on the contextual dependency in the temporal domain and neglect the spatial configurations of articulated skeletons. In this paper, we propose a novel two-stream RNN architecture to model both temporal dynamics and spatial configurations for skeleton based action recognition. We explore two different structures for the temporal stream: stacked RNN and hierarchical RNN. Hierarchical RNN is designed according to human body kinematics. We also propose two effective methods to model the spatial structure by converting the spatial graph into a sequence of joints. To improve generalization of our model, we further exploit 3D transformation based data augmentation techniques including rotation and scaling transformation to transform the 3D coordinates of skeletons during training. Experiments on 3D action recognition benchmark datasets show that our method brings a considerable improvement for a variety of actions, i.e., generic actions, interaction activities and gestures.",
"Recently, skeleton based action recognition gains more popularity due to cost-effective depth sensors coupled with real-time skeleton estimation algorithms. Traditional approaches based on handcrafted features are limited to represent the complexity of motion patterns. Recent methods that use Recurrent Neural Networks (RNN) to handle raw skeletons only focus on the contextual dependency in the temporal domain and neglect the spatial configurations of articulated skeletons. In this paper, we propose a novel two-stream RNN architecture to model both temporal dynamics and spatial configurations for skeleton based action recognition. We explore two different structures for the temporal stream: stacked RNN and hierarchical RNN. Hierarchical RNN is designed according to human body kinematics. We also propose two effective methods to model the spatial structure by converting the spatial graph into a sequence of joints. To improve generalization of our model, we further exploit 3D transformation based data augmentation techniques including rotation and scaling transformation to transform the 3D coordinates of skeletons during training. Experiments on 3D action recognition benchmark datasets show that our method brings a considerable improvement for a variety of actions, i.e., generic actions, interaction activities and gestures.",
"In recent years, the convolutional neural network (CNN) has achieved great success in many computer vision tasks. Partially inspired by neuroscience, CNN shares many properties with the visual system of the brain. A prominent difference is that CNN is typically a feed-forward architecture while in the visual system recurrent connections are abundant. Inspired by this fact, we propose a recurrent CNN (RCNN) for object recognition by incorporating recurrent connections into each convolutional layer. Though the input is static, the activities of RCNN units evolve over time so that the activity of each unit is modulated by the activities of its neighboring units. This property enhances the ability of the model to integrate the context information, which is important for object recognition. Like other recurrent neural networks, unfolding the RCNN through time can result in an arbitrarily deep network with a fixed number of parameters. Furthermore, the unfolded network has multiple paths, which can facilitate the learning process. The model is tested on four benchmark object recognition datasets: CIFAR-10, CIFAR-100, MNIST and SVHN. With fewer trainable parameters, RCNN outperforms the state-of-the-art models on all of these datasets. Increasing the number of parameters leads to even better performance. These results demonstrate the advantage of the recurrent structure over purely feed-forward structure for object recognition.",
"Human actions can be represented by the trajectories of skeleton joints. Traditional methods generally model the spatial structure and temporal dynamics of human skeleton with hand-crafted features and recognize human actions by well-designed classifiers. In this paper, considering that recurrent neural network (RNN) can model the long-term contextual information of temporal sequences well, we propose an end-to-end hierarchical RNN for skeleton based action recognition. Instead of taking the whole skeleton as the input, we divide the human skeleton into five parts according to human physical structure, and then separately feed them to five subnets. As the number of layers increases, the representations extracted by the subnets are hierarchically fused to be the inputs of higher layers. The final representations of the skeleton sequences are fed into a single-layer perceptron, and the temporally accumulated output of the perceptron is the final decision. We compare with five other deep RNN architectures derived from our model to verify the effectiveness of the proposed network, and also compare with several other methods on three publicly available datasets. Experimental results demonstrate that our model achieves the state-of-the-art performance with high computational efficiency."
]
} |
1907.09682 | 2964161024 | Knowledge distillation is a widely applicable technique for training a student neural network under the guidance of a trained teacher network. For example, in neural network compression, a high-capacity teacher is distilled to train a compact student; in privileged learning, a teacher trained with privileged data is distilled to train a student without access to that data. The distillation loss determines how a teacher's knowledge is captured and transferred to the student. In this paper, we propose a new form of knowledge distillation loss that is inspired by the observation that semantically similar inputs tend to elicit similar activation patterns in a trained network. Similarity-preserving knowledge distillation guides the training of a student network such that input pairs that produce similar (dissimilar) activations in the teacher network produce similar (dissimilar) activations in the student network. In contrast to previous distillation methods, the student is not required to mimic the representation space of the teacher, but rather to preserve the pairwise similarities in its own representation space. Experiments on three public datasets demonstrate the potential of our approach. | We presented in this paper a novel distillation loss for capturing and transferring knowledge from a teacher network to a student network. Several prior alternatives @cite_44 @cite_36 @cite_45 @cite_14 are described in the introduction and some key differences are highlighted in Section . In addition to the knowledge capture (or loss definition) aspect of distillation studied in this paper, another important open question is the architectural design of students and teachers. In most studies of knowledge distillation, including ours, the student network is a thinner and or shallower version of the teacher network. Inspired by efficient architectures such as MobileNet and ShuffleNet, @cite_16 proposed to replace regular convolutions in the teacher network with cheaper grouped and pointwise convolutions in the student. @cite_29 developed a reinforcement learning approach to learn the student architecture. @cite_18 demonstrated how a quantized student network can be trained using a full-precision teacher network. | {
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_36",
"@cite_29",
"@cite_44",
"@cite_45",
"@cite_16"
],
"mid": [
"2951168573",
"2803113663",
"2807912816",
"2620998106",
"2962935923",
"2895977466",
"2787752464"
],
"abstract": [
"Model distillation is an effective and widely used technique to transfer knowledge from a teacher to a student network. The typical application is to transfer from a powerful large network or ensemble to a small network, that is better suited to low-memory or fast execution requirements. In this paper, we present a deep mutual learning (DML) strategy where, rather than one way transfer between a static pre-defined teacher and a student, an ensemble of students learn collaboratively and teach each other throughout the training process. Our experiments show that a variety of network architectures benefit from mutual learning and achieve compelling results on CIFAR-100 recognition and Market-1501 person re-identification benchmarks. Surprisingly, it is revealed that no prior powerful teacher network is necessary -- mutual learning of a collection of simple student networks works, and moreover outperforms distillation from a more powerful yet static teacher.",
"In this paper, we propose an efficient and fast object detector which can process hundreds of frames per second. To achieve this goal we investigate three main aspects of the object detection framework: network architecture, loss function and training data (labeled and unlabeled). In order to obtain compact network architecture, we introduce various improvements, based on recent work, to develop an architecture which is computationally light-weight and achieves a reasonable performance. To further improve the performance, while keeping the complexity same, we utilize distillation loss function. Using distillation loss we transfer the knowledge of a more accurate teacher network to proposed light-weight student network. We propose various innovations to make distillation efficient for the proposed one stage detector pipeline: objectness scaled distillation loss, feature map non-maximal suppression and a single unified distillation loss function for detection. Finally, building upon the distillation loss, we explore how much can we push the performance by utilizing the unlabeled data. We train our model with unlabeled data using the soft labels of the teacher network. Our final network consists of 10x fewer parameters than the VGG based object detection network and it achieves a speed of more than 200 FPS and proposed changes improve the detection accuracy by 14 mAP over the baseline on Pascal dataset.",
"Knowledge distillation is effective to train small and generalisable network models for meeting the low-memory and fast running requirements. Existing offline distillation methods rely on a strong pre-trained teacher, which enables favourable knowledge discovery and transfer but requires a complex two-phase training procedure. Online counterparts address this limitation at the price of lacking a highcapacity teacher. In this work, we present an On-the-fly Native Ensemble (ONE) strategy for one-stage online distillation. Specifically, ONE trains only a single multi-branch network while simultaneously establishing a strong teacher on-the- fly to enhance the learning of target network. Extensive evaluations show that ONE improves the generalisation performance a variety of deep neural networks more significantly than alternative methods on four image classification dataset: CIFAR10, CIFAR100, SVHN, and ImageNet, whilst having the computational efficiency advantages.",
"Model distillation is an effective and widely used technique to transfer knowledge from a teacher to a student network. The typical application is to transfer from a powerful large network or ensemble to a small network, in order to meet the low-memory or fast execution requirements. In this paper, we present a deep mutual learning (DML) strategy. Different from the one-way transfer between a static pre-defined teacher and a student in model distillation, with DML, an ensemble of students learn collaboratively and teach each other throughout the training process. Our experiments show that a variety of network architectures benefit from mutual learning and achieve compelling results on both category and instance recognition tasks. Surprisingly, it is revealed that no prior powerful teacher network is necessary - mutual learning of a collection of simple student networks works, and moreover outperforms distillation from a more powerful yet static teacher.",
"Many engineers wish to deploy modern neural networks in memory-limited settings; but the development of flexible methods for reducing memory use is in its infancy, and there is little knowledge of the resulting cost-benefit. We propose structural model distillation for memory reduction using a strategy that produces a student architecture that is a simple transformation of the teacher architecture: no redesign is needed, and the same hyperparameters can be used. Using attention transfer, we provide Pareto curves tables for distillation of residual networks with four benchmark datasets, indicating the memory versus accuracy payoff. We show that substantial memory savings are possible with very little loss of accuracy, and confirm that distillation provides student network performance that is better than training that student architecture directly on data.",
"To reduce the large computation and storage cost of a deep convolutional neural network, the knowledge distillation based methods have pioneered to transfer the generalization ability of a large (teacher) deep network to a light-weight (student) network. However, these methods mostly focus on transferring the probability distribution of the softmax layer in a teacher network and thus neglect the intermediate representations. In this paper, we propose a knowledge transfer adversarial network to better train a student network. Our technique holistically considers both intermediate representations and probability distributions of a teacher network. To transfer the knowledge of intermediate representations, we set high-level teacher feature maps as a target, toward which the student feature maps are trained. Specifically, we arrange a Teacher-to-Student layer for enabling our framework suitable for various student structures. The intermediate representation helps the student network better understand the transferred generalization as compared to the probability distribution only. Furthermore, we infuse an adversarial learning process by employing a discriminator network, which can fully exploit the spatial correlation of feature maps in training a student network. The experimental results demonstrate that the proposed method can significantly improve the performance of a student network on both image classification and object detection tasks.",
"Deep neural networks (DNNs) continue to make significant advances, solving tasks from image classification to translation or reinforcement learning. One aspect of the field receiving considerable attention is efficiently executing deep models in resource-constrained environments, such as mobile or embedded devices. This paper focuses on this problem, and proposes two new compression methods, which jointly leverage weight quantization and distillation of larger teacher networks into smaller student networks. The first method we propose is called quantized distillation and leverages distillation during the training process, by incorporating distillation loss, expressed with respect to the teacher, into the training of a student network whose weights are quantized to a limited set of levels. The second method, differentiable quantization, optimizes the location of quantization points through stochastic gradient descent, to better fit the behavior of the teacher model. We validate both methods through experiments on convolutional and recurrent architectures. We show that quantized shallow students can reach similar accuracy levels to full-precision teacher models, while providing order of magnitude compression, and inference speedup that is linear in the depth reduction. In sum, our results enable DNNs for resource-constrained environments to leverage architecture and accuracy advances developed on more powerful devices."
]
} |
1907.09682 | 2964161024 | Knowledge distillation is a widely applicable technique for training a student neural network under the guidance of a trained teacher network. For example, in neural network compression, a high-capacity teacher is distilled to train a compact student; in privileged learning, a teacher trained with privileged data is distilled to train a student without access to that data. The distillation loss determines how a teacher's knowledge is captured and transferred to the student. In this paper, we propose a new form of knowledge distillation loss that is inspired by the observation that semantically similar inputs tend to elicit similar activation patterns in a trained network. Similarity-preserving knowledge distillation guides the training of a student network such that input pairs that produce similar (dissimilar) activations in the teacher network produce similar (dissimilar) activations in the student network. In contrast to previous distillation methods, the student is not required to mimic the representation space of the teacher, but rather to preserve the pairwise similarities in its own representation space. Experiments on three public datasets demonstrate the potential of our approach. | State-of-the-art network compression methods can achieve significant reductions in network size, in some cases by an order of magnitude, but often require specialized software or hardware support. For example, unstructured pruning requires optimized sparse matrix multiplication routines to realize practical acceleration @cite_32 , platform constraint-aware compression @cite_35 @cite_6 @cite_15 requires hardware simulators or empirical measurements, and arbitrary-bit quantization @cite_42 @cite_17 requires specialized hardware. One of the advantages of knowledge distillation is that it is easily implemented in any off-the-shelf deep learning framework without the need for extra software or hardware. Moreover, distillation can be integrated with other network compression techniques for further gains in performance @cite_18 . | {
"cite_N": [
"@cite_35",
"@cite_18",
"@cite_42",
"@cite_32",
"@cite_6",
"@cite_15",
"@cite_17"
],
"mid": [
"2915589364",
"2963723401",
"2806364818",
"2788715907",
"2767012653",
"2895260923",
"2526792573"
],
"abstract": [
"We rigorously evaluate three state-of-the-art techniques for inducing sparsity in deep neural networks on two large-scale learning tasks: Transformer trained on WMT 2014 English-to-German, and ResNet-50 trained on ImageNet. Across thousands of experiments, we demonstrate that complex techniques (, 2017; , 2017b) shown to yield high compression rates on smaller datasets perform inconsistently, and that simple magnitude pruning approaches achieve comparable or better results. Additionally, we replicate the experiments performed by (Frankle & Carbin, 2018) and (, 2018) at scale and show that unstructured sparse architectures learned through pruning cannot be trained from scratch to the same test set performance as a model trained with joint sparsification and optimization. Together, these results highlight the need for large-scale benchmarks in the field of model compression. We open-source our code, top performing model checkpoints, and results of all hyperparameter configurations to establish rigorous baselines for future work on compression and sparsification.",
"Deep learning networks have achieved state-of-the-art accuracies on computer vision workloads like image classification and object detection. The performant systems, however, typically involve big models with numerous parameters. Once trained, a challenging aspect for such top performing models is deployment on resource constrained inference systems -- the models (often deep networks or wide networks or both) are compute and memory intensive. Low precision numerics and model compression using knowledge distillation are popular techniques to lower both the compute requirements and memory footprint of these deployed models. In this paper, we study the combination of these two techniques and show that the performance of low precision networks can be significantly improved by using knowledge distillation techniques. We call our approach Apprentice and show state-of-the-art accuracies using ternary precision and 4-bit precision for many variants of ResNet architecture on ImageNet dataset. We study three schemes in which one can apply knowledge distillation techniques to various stages of the train-and-deploy pipeline.",
"Deep neural networks (DNNs) have become the state-of-the-art technique for machine learning tasks in various applications. However, due to their size and the computational complexity, large DNNs are not readily deployable on edge devices in real-time. To manage complexity and accelerate computation, network compression techniques based on pruning and quantization have been proposed and shown to be effective in reducing network size. However, such network compression can result in irregular matrix structures that are mismatched with modern hardware-accelerated platforms, such as graphics processing units (GPUs) designed to perform the DNN matrix multiplications in a structured (block-based) way. We propose MPDCompress, a DNN compression algorithm based on matrix permutation decomposition via random mask generation. In-training application of the masks molds the synaptic weight connection matrix to a sub-graph separation format. Aided by the random permutations, a hardware-desirable block matrix is generated, allowing for a more efficient implementation and compression of the network. To show versatility, we empirically verify MPDCompress on several network models, compression rates, and image datasets. On the LeNet 300-100 model (MNIST dataset), Deep MNIST, and CIFAR10, we achieve 10 X network compression with less than 1 accuracy loss compared to non-compressed accuracy performance. On AlexNet for the full ImageNet ILSVRC-2012 dataset, we achieve 8 X network compression with less than 1 accuracy loss, with top-5 and top-1 accuracies of 79.6 and 56.4 , respectively. Finally, we observe that the algorithm can offer inference speedups across various hardware platforms, with 4 X faster operation achieved on several mobile GPUs.",
"In recent years considerable research efforts have been devoted to compression techniques of convolutional neural networks (CNNs). Many works so far have focused on CNN connection pruning methods which produce sparse parameter tensors in convolutional or fully-connected layers. It has been demonstrated in several studies that even simple methods can effectively eliminate connections of a CNN. However, since these methods make parameter tensors just sparser but no smaller, the compression may not transfer directly to acceleration without support from specially designed hardware. In this paper, we propose an iterative approach named Auto-balanced Filter Pruning, where we pre-train the network in an innovative auto-balanced way to transfer the representational capacity of its convolutional layers to a fraction of the filters, prune the redundant ones, then re-train it to restore the accuracy. In this way, a smaller version of the original network is learned and the floating-point operations (FLOPs) are reduced. By applying this method on several common CNNs, we show that a large portion of the filters can be discarded without obvious accuracy drop, leading to significant reduction of computational burdens. Concretely, we reduce the inference cost of LeNet-5 on MNIST, VGG-16 and ResNet-56 on CIFAR-10 by 95.1 , 79.7 and 60.9 , respectively. Copyright © 2018, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.",
"Model compression and knowledge distillation have been successfully applied for cross-architecture and cross-domain transfer learning. However, a key requirement is that training examples are in correspondence across the domains. We show that in many scenarios of practical importance such aligned data can be synthetically generated using computer graphics pipelines allowing domain adaptation through distillation. We apply this technique to learn models for recognizing low-resolution images using labeled high-resolution images, non-localized objects using labeled localized objects, line-drawings using labeled color images, etc. Experiments on various fine-grained recognition datasets demonstrate that the technique improves recognition performance on the low-quality data and beats strong baselines for domain adaptation. Finally, we present insights into workings of the technique through visualizations and relating it to existing literature.",
"Deep neural network compression has the potential to bring modern resource-hungry deep networks to resource-limited devices. However, in many of the most compelling deployment scenarios of compressed deep networks, the operational constraints matter: for example, a pedestrian detection network on a self-driving car may have to satisfy a latency constraint for safe operation. We propose the first principled treatment of deep network compression under operational constraints. We formulate the compression learning problem from the perspective of constrained Bayesian optimization, and introduce a cooling (annealing) strategy to guide the network compression towards the target constraints. Experiments on ImageNet demonstrate the value of modelling constraints directly in network compression.",
"Lossy compression introduces complex compression artifacts, particularly blocking artifacts, ringing effects and blurring. Existing algorithms either focus on removing blocking artifacts and produce blurred output, or restore sharpened images that are accompanied with ringing effects. Inspired by the success of deep convolutional networks (DCN) on superresolution, we formulate a compact and efficient network for seamless attenuation of different compression artifacts. To meet the speed requirement of real-world applications, we further accelerate the proposed baseline model by layer decomposition and joint use of large-stride convolutional and deconvolutional layers. This also leads to a more general CNN framework that has a close relationship with the conventional Multi-Layer Perceptron (MLP). Finally, the modified network achieves a speed up of 7.5 times with almost no performance loss compared to the baseline model. We also demonstrate that a deeper model can be effectively trained with features learned in a shallow network. Following a similar \"easy to hard\" idea, we systematically investigate three practical transfer settings and show the effectiveness of transfer learning in low-level vision problems. Our method shows superior performance than the state-of-the-art methods both on benchmark datasets and a real-world use case."
]
} |
1907.09702 | 2963448607 | Temporal action proposal generation is an challenging and promising task which aims to locate temporal regions in real-world videos where action or event may occur. Current bottom-up proposal generation methods can generate proposals with precise boundary, but cannot efficiently generate adequately reliable confidence scores for retrieving proposals. To address these difficulties, we introduce the Boundary-Matching (BM) mechanism to evaluate confidence scores of densely distributed proposals, which denote a proposal as a matching pair of starting and ending boundaries and combine all densely distributed BM pairs into the BM confidence map. Based on BM mechanism, we propose an effective, efficient and end-to-end proposal generation method, named Boundary-Matching Network (BMN), which generates proposals with precise temporal boundaries as well as reliable confidence scores simultaneously. The two-branches of BMN are jointly trained in an unified framework. We conduct experiments on two challenging datasets: THUMOS-14 and ActivityNet-1.3, where BMN shows significant performance improvement with remarkable efficiency and generalizability. Further, combining with existing action classifier, BMN can achieve state-of-the-art temporal action detection performance. | Action recognition is a fundamental and important task of video understanding area. Hand-crafted features such as HOG, HOF and MBH are widely used in earlier works, such as improved Dense Trajectory (iDT) @cite_21 @cite_30 . Recently, deep learning models have achieved significantly performance promotion in action recognition task. The mainstream networks fall into two categories: two-stream networks @cite_19 @cite_34 @cite_22 exploit appearance and motion clues from RGB image and stacked optical flow separately; 3D networks @cite_14 @cite_6 exploit appearance and motion clues directly from raw video volume. In our work, by convention, we adopt action recognition models to extract visual feature sequence of untrimmed video. | {
"cite_N": [
"@cite_30",
"@cite_14",
"@cite_22",
"@cite_21",
"@cite_6",
"@cite_19",
"@cite_34"
],
"mid": [
"2472293097",
"2057815930",
"1981781955",
"2131042978",
"2609658199",
"2962778061",
"1871385855"
],
"abstract": [
"Recently, deep learning approaches have demonstrated remarkable progresses for action recognition in videos. Most existing deep frameworks equally treat every volume i.e. spatial-temporal video clip, and directly assign a video label to all volumes sampled from it. However, within a video, discriminative actions may occur sparsely in a few key volumes, and most other volumes are irrelevant to the labeled action category. Training with a large proportion of irrelevant volumes will hurt performance. To address this issue, we propose a key volume mining deep framework to identify key volumes and conduct classification simultaneously. Specifically, our framework is trained is optimized in an alternative way integrated to the forward and backward stages of Stochastic Gradient Descent (SGD). In the forward pass, our network mines key volumes for each action class. In the backward pass, it updates network parameters with the help of these mined key volumes. In addition, we propose \"Stochastic out\" to model key volumes from multi-modalities, and an effective yet simple \"unsupervised key volume proposal\" method for high quality volume sampling. Our experiments show that action recognition performance can be significantly improved by mining key volumes, and we achieve state-of-the-art performance on HMDB51 and UCF101 (93.1 ).",
"Most of the previous work on video action recognition use complex hand-designed local features, such as SIFT, HOG and SURF, but these approaches are implemented sophisticatedly and difficult to be extended to other sensor modalities. Recent studies discover that there are no universally best hand-engineered features for all datasets, and learning features directly from the data may be more advantageous. One such endeavor is Slow Feature Analysis (SFA) proposed by Wiskott and Sejnowski [33]. SFA can learn the invariant and slowly varying features from input signals and has been proved to be valuable in human action recognition [34]. It is also observed that the multi-layer feature representation has succeeded remarkably in widespread machine learning applications. In this paper, we propose to combine SFA with deep learning techniques to learn hierarchical representations from the video data itself. Specifically, we use a two-layered SFA learning structure with 3D convolution and max pooling operations to scale up the method to large inputs and capture abstract and structural features from the video. Thus, the proposed method is suitable for action recognition. At the same time, sharing the same merits of deep learning, the proposed method is generic and fully automated. Our classification results on Hollywood2, KTH and UCF Sports are competitive with previously published results. To highlight some, on the KTH dataset, our recognition rate shows approximately 1 improvement in comparison to state-of-the-art methods even without supervision or dense sampling.",
"Action recognition on large categories of unconstrained videos taken from the web is a very challenging problem compared to datasets like KTH (6 actions), IXMAS (13 actions), and Weizmann (10 actions). Challenges like camera motion, different viewpoints, large interclass variations, cluttered background, occlusions, bad illumination conditions, and poor quality of web videos cause the majority of the state-of-the-art action recognition approaches to fail. Also, an increased number of categories and the inclusion of actions with high confusion add to the challenges. In this paper, we propose using the scene context information obtained from moving and stationary pixels in the key frames, in conjunction with motion features, to solve the action recognition problem on a large (50 actions) dataset with videos from the web. We perform a combination of early and late fusion on multiple features to handle the very large number of categories. We demonstrate that scene context is a very important feature to perform action recognition on very large datasets. The proposed method does not require any kind of video stabilization, person detection, or tracking and pruning of features. Our approach gives good performance on a large number of action categories; it has been tested on the UCF50 dataset with 50 action categories, which is an extension of the UCF YouTube Action (UCF11) dataset containing 11 action categories. We also tested our approach on the KTH and HMDB51 datasets for comparison.",
"Action recognition in uncontrolled video is an important and challenging computer vision problem. Recent progress in this area is due to new local features and models that capture spatio-temporal structure between local features, or human-object interactions. Instead of working towards more complex models, we focus on the low-level features and their encoding. We evaluate the use of Fisher vectors as an alternative to bag-of-word histograms to aggregate a small set of state-of-the-art low-level descriptors, in combination with linear classifiers. We present a large and varied set of evaluations, considering (i) classification of short actions in five datasets, (ii) localization of such actions in feature-length movies, and (iii) large-scale recognition of complex events. We find that for basic action recognition and localization MBH features alone are enough for state-of-the-art performance. For complex events we find that SIFT and MFCC features provide complementary cues. On all three problems we obtain state-of-the-art results, while using fewer features and less complex models.",
"Suitable feature representation is essential for performing video analysis and understanding in applications within the smart surveillance domain. In this paper, we propose a novel spatiotemporal feature descriptor based on co-occurrence matrices computed from the optical flow magnitude and orientation. Our method, called Optical Flow Co-occurrence Matrices (OFCM), extracts a robust set of measures known as Haralick features to describe the flow patterns by measuring meaningful properties such as contrast, entropy and homogeneity of co-occurrence matrices to capture local space-time characteristics of the motion through the neighboring optical flow magnitude and orientation. We evaluate the proposed method on the action recognition problem by applying a visual recognition pipeline involving bag of local spatiotemporal features and SVM classification. The experimental results, carried on three well-known datasets (KTH, UCF Sports and HMDB51), demonstrate that OFCM outperforms the results achieved by several widely employed spatiotemporal feature descriptors such as HOF, HOG3D and MBH, indicating its suitability to be used as video representation.",
"Deep learning for human action recognition in videos is making significant progress, but is slowed down by its dependency on expensive manual labeling of large video collections. In this work, we investigate the generation of synthetic training data for action recognition, as it has recently shown promising results for a variety of other computer vision tasks. We propose an interpretable parametric generative model of human action videos that relies on procedural generation and other computer graphics techniques of modern game engines. We generate a diverse, realistic, and physically plausible dataset of human action videos, called PHAV for Procedural Human Action Videos. It contains a total of 39,982 videos, with more than 1,000 examples for each action of 35 categories. Our approach is not limited to existing motion capture sequences, and we procedurally define 14 synthetic actions. We introduce a deep multi-task representation learning architecture to mix synthetic and real videos, even if the action categories differ. Our experiments on the UCF101 and HMDB51 benchmarks suggest that combining our large set of synthetic videos with small real-world datasets can boost recognition performance, significantly outperforming fine-tuning state-of-the-art unsupervised generative models of videos.",
"A Comprehensive study on the BoVW pipeline for action recognition task.An evaluation and a generic analysis on 13 encoding methods.Intra-normalization for supervector based encoding methods.An evaluation on three fusion methods.Several good practices of the BoVW pipeline for action recognition task. Video based action recognition is one of the important and challenging problems in computer vision research. Bag of visual words model (BoVW) with local features has been very popular for a long time and obtained the state-of-the-art performance on several realistic datasets, such as the HMDB51, UCF50, and UCF101. BoVW is a general pipeline to construct a global representation from local features, which is mainly composed of five steps; (i) feature extraction, (ii) feature pre-processing, (iii) codebook generation, (iv) feature encoding, and (v) pooling and normalization. Although many efforts have been made in each step independently in different scenarios, their effects on action recognition are still unknown. Meanwhile, video data exhibits different views of visual patterns , such as static appearance and motion dynamics. Multiple descriptors are usually extracted to represent these different views. Fusing these descriptors is crucial for boosting the final performance of an action recognition system. This paper aims to provide a comprehensive study of all steps in BoVW and different fusion methods, and uncover some good practices to produce a state-of-the-art action recognition system. Specifically, we explore two kinds of local features, ten kinds of encoding methods, eight kinds of pooling and normalization strategies, and three kinds of fusion methods. We conclude that every step is crucial for contributing to the final recognition rate and improper choice in one of the steps may counteract the performance improvement of other steps. Furthermore, based on our comprehensive study, we propose a simple yet effective representation, called hybrid supervector, by exploring the complementarity of different BoVW frameworks with improved dense trajectories. Using this representation, we obtain impressive results on the three challenging datasets; HMDB51 (61.9 ), UCF50 (92.3 ), and UCF101 (87.9 )."
]
} |
1907.09702 | 2963448607 | Temporal action proposal generation is an challenging and promising task which aims to locate temporal regions in real-world videos where action or event may occur. Current bottom-up proposal generation methods can generate proposals with precise boundary, but cannot efficiently generate adequately reliable confidence scores for retrieving proposals. To address these difficulties, we introduce the Boundary-Matching (BM) mechanism to evaluate confidence scores of densely distributed proposals, which denote a proposal as a matching pair of starting and ending boundaries and combine all densely distributed BM pairs into the BM confidence map. Based on BM mechanism, we propose an effective, efficient and end-to-end proposal generation method, named Boundary-Matching Network (BMN), which generates proposals with precise temporal boundaries as well as reliable confidence scores simultaneously. The two-branches of BMN are jointly trained in an unified framework. We conduct experiments on two challenging datasets: THUMOS-14 and ActivityNet-1.3, where BMN shows significant performance improvement with remarkable efficiency and generalizability. Further, combining with existing action classifier, BMN can achieve state-of-the-art temporal action detection performance. | Correlation matching algorithms are widely used in many computer vision tasks, such as image registration, action recognition and stereo matching. Specifically, stereo matching aims to find corresponding pixels from stereo images. For each pixel in left image of a rectified image pair, the stereo matching method need to find corresponding pixel in right image along horizontal direction, or we can say finding right pixel with minimum cost. Thus, the cost minimization of all left pixels can be denoted as a cost volume, which denotes each left-right pixel pair as a point in volume. Based on cost volume, many recent works @cite_9 @cite_36 @cite_31 achieve end-to-end network via generating cost volume directly from combining two feature maps, using correlation layer @cite_36 or feature concatenation @cite_3 . Inspired by cost volume, our proposed BM confidence map contains pairs of temporal starting and ending boundaries as proposals, thus can directly generate confidence scores for all proposals using convolutional layers. We propose BM layer to efficiently generate BM feature map via sampling feature among starting and ending boundaries of each proposal simultaneously. | {
"cite_N": [
"@cite_36",
"@cite_9",
"@cite_31",
"@cite_3"
],
"mid": [
"1772650917",
"2963502507",
"2144041313",
"2113873920"
],
"abstract": [
"We present a method for extracting depth information from a rectified image pair. Our approach focuses on the first stage of many stereo algorithms: the matching cost computation. We approach the problem by learning a similarity measure on small image patches using a convolutional neural network. Training is carried out in a supervised manner by constructing a binary classification data set with examples of similar and dissimilar pairs of patches. We examine two network architectures for this task: one tuned for speed, the other for accuracy. The output of the convolutional neural network is used to initialize the stereo matching cost. A series of post-processing steps follow: cross-based cost aggregation, semiglobal matching, a left-right consistency check, subpixel enhancement, a median filter, and a bilateral filter. We evaluate our method on the KITTI 2012, KITTI 2015, and Middlebury stereo data sets and show that it outperforms other approaches on all three data sets.",
"We present a method for extracting depth information from a rectified image pair. Our approach focuses on the first stage of many stereo algorithms: the matching cost computation. We approach the problem by learning a similarity measure on small image patches using a convolutional neural network. Training is carried out in a supervised manner by constructing a binary classification data set with examples of similar and dissimilar pairs of patches. We examine two network architectures for this task: one tuned for speed, the other for accuracy. The output of the convolutional neural network is used to initialize the stereo matching cost. A series of post-processing steps follow: cross-based cost aggregation, semiglobal matching, a left-right consistency check, subpixel enhancement, a median filter, and a bilateral filter. We evaluate our method on the KITTI 2012, KITTI 2015, and Middlebury stereo data sets and show that it outperforms other approaches on all three data sets.",
"We present a method for extracting depth information from a rectified image pair. We train a convolutional neural network to predict how well two image patches match and use it to compute the stereo matching cost. The cost is refined by cross-based cost aggregation and semiglobal matching, followed by a left-right consistency check to eliminate errors in the occluded regions. Our stereo method achieves an error rate of 2.61 on the KITTI stereo dataset and is currently (August 2014) the top performing method on this dataset.",
"We propose an area-based local stereo matching algorithm for accurate disparity estimation across all image regions. A well-known challenge to local stereo methods is to decide an appropriate support window for the pixel under consideration, adapting the window shape or the pixelwise support weight to the underlying scene structures. Our stereo method tackles this problem with two key contributions. First, for each anchor pixel an upright cross local support skeleton is adaptively constructed, with four varying arm lengths decided on color similarity and connectivity constraints. Second, given the local cross-decision results, we dynamically construct a shape-adaptive full support region on the fly, merging horizontal segments of the crosses in the vertical neighborhood. Approximating image structures accurately, the proposed method is among the best performing local stereo methods according to the benchmark Middlebury stereo evaluation. Additionally, it reduces memory consumption significantly thanks to our compact local cross representation. To accelerate matching cost aggregation performed in an arbitrarily shaped 2-D region, we also propose an orthogonal integral image technique, yielding a speedup factor of 5-15 over the straightforward integration."
]
} |
1907.09702 | 2963448607 | Temporal action proposal generation is an challenging and promising task which aims to locate temporal regions in real-world videos where action or event may occur. Current bottom-up proposal generation methods can generate proposals with precise boundary, but cannot efficiently generate adequately reliable confidence scores for retrieving proposals. To address these difficulties, we introduce the Boundary-Matching (BM) mechanism to evaluate confidence scores of densely distributed proposals, which denote a proposal as a matching pair of starting and ending boundaries and combine all densely distributed BM pairs into the BM confidence map. Based on BM mechanism, we propose an effective, efficient and end-to-end proposal generation method, named Boundary-Matching Network (BMN), which generates proposals with precise temporal boundaries as well as reliable confidence scores simultaneously. The two-branches of BMN are jointly trained in an unified framework. We conduct experiments on two challenging datasets: THUMOS-14 and ActivityNet-1.3, where BMN shows significant performance improvement with remarkable efficiency and generalizability. Further, combining with existing action classifier, BMN can achieve state-of-the-art temporal action detection performance. | As aforementioned, the goal of temporal action detection task is to detect action instances in untrimmed videos with temporal boundaries and action categories, which can be divided into temporal proposal generation and action classification stages. These two stages are taken apart in most detection methods @cite_2 @cite_24 @cite_28 , and are taken together as single model in some methods @cite_35 @cite_27 . For proposal generation task, most previous works @cite_10 @cite_23 @cite_4 @cite_16 @cite_2 adopt fashion to generate proposals with pre-defined duration and interval, where the main drawback is the lack of boundary precision and duration flexibility. There are also some methods @cite_28 @cite_26 adopt fashion. TAG @cite_28 generates proposals using temporal watershed algorithm, but lack confidence scores for retrieving. Recently, BSN @cite_26 generates proposals via locally locating temporal boundaries and globally evaluating confidence scores, and achieves significant performance promotion over previous proposal generation methods. In this work, we propose the Boundary-Matching mechanism for proposal confidence evaluation, which can largely simplify the pipeline of BSN and bring significant promotion in both efficiency and effectiveness. | {
"cite_N": [
"@cite_35",
"@cite_26",
"@cite_4",
"@cite_28",
"@cite_24",
"@cite_27",
"@cite_23",
"@cite_2",
"@cite_16",
"@cite_10"
],
"mid": [
"2962677524",
"2805042136",
"2766402183",
"2895738954",
"2670080482",
"2607644813",
"2964216549",
"2484328966",
"2949310145",
"2951473326"
],
"abstract": [
"Temporal action proposal generation is an important yet challenging problem, since temporal proposals with rich action content are indispensable for analysing real-world videos with long duration and high proportion irrelevant content. This problem requires methods not only generating proposals with precise temporal boundaries, but also retrieving proposals to cover truth action instances with high recall and high overlap using relatively fewer proposals. To address these difficulties, we introduce an effective proposal generation method, named Boundary-Sensitive Network (BSN), which adopts “local to global” fashion. Locally, BSN first locates temporal boundaries with high probabilities, then directly combines these boundaries as proposals. Globally, with Boundary-Sensitive Proposal feature, BSN retrieves proposals by evaluating the confidence of whether a proposal contains an action within its region. We conduct experiments on two challenging datasets: ActivityNet-1.3 and THUMOS14, where BSN outperforms other state-of-the-art temporal action proposal generation methods with high recall and high temporal precision. Finally, further experiments demonstrate that by combining existing action classifiers, our method significantly improves the state-of-the-art temporal action detection performance.",
"Temporal action proposal generation is an important yet challenging problem, since temporal proposals with rich action content are indispensable for analysing real-world videos with long duration and high proportion irrelevant content. This problem requires methods not only generating proposals with precise temporal boundaries, but also retrieving proposals to cover truth action instances with high recall and high overlap using relatively fewer proposals. To address these difficulties, we introduce an effective proposal generation method, named Boundary-Sensitive Network (BSN), which adopts \"local to global\" fashion. Locally, BSN first locates temporal boundaries with high probabilities, then directly combines these boundaries as proposals. Globally, with Boundary-Sensitive Proposal feature, BSN retrieves proposals by evaluating the confidence of whether a proposal contains an action within its region. We conduct experiments on two challenging datasets: ActivityNet-1.3 and THUMOS14, where BSN outperforms other state-of-the-art temporal action proposal generation methods with high recall and high temporal precision. Finally, further experiments demonstrate that by combining existing action classifiers, our method significantly improves the state-of-the-art temporal action detection performance.",
"Temporal action detection is a very important yet challenging problem, since videos in real applications are usually long, untrimmed and contain multiple action instances. This problem requires not only recognizing action categories but also detecting start time and end time of each action instance. Many state-of-the-art methods adopt the \"detection by classification\" framework: first do proposal, and then classify proposals. The main drawback of this framework is that the boundaries of action instance proposals have been fixed during the classification step. To address this issue, we propose a novel Single Shot Action Detector (SSAD) network based on 1D temporal convolutional layers to skip the proposal generation step via directly detecting action instances in untrimmed video. On pursuit of designing a particular SSAD network that can work effectively for temporal action detection, we empirically search for the best network architecture of SSAD due to lacking existing models that can be directly adopted. Moreover, we investigate into input feature types and fusion strategies to further improve detection accuracy. We conduct extensive experiments on two challenging datasets: THUMOS 2014 and MEXaction2. When setting Intersection-over-Union threshold to 0.5 during evaluation, SSAD significantly outperforms other state-of-the-art systems by increasing mAP from @math to @math on THUMOS 2014 and from 7.4 to @math on MEXaction2.",
"Detecting actions in videos is a challenging task as video is an information intensive media with complex variations. Existing approaches predominantly generate action proposals for each individual frame or fixed-length clip independently, while overlooking temporal context across them. Such temporal contextual relations are vital for action detection as an action is by nature a sequence of movements. This motivates us to leverage the localized action proposals in previous frames when determining action regions in the current one. Specifically, we present a novel deep architecture called Recurrent Tubelet Proposal and Recognition (RTPR) networks to incorporate temporal context for action detection. The proposed RTPR consists of two correlated networks, i.e., Recurrent Tubelet Proposal (RTP) networks and Recurrent Tubelet Recognition (RTR) networks. The RTP initializes action proposals of the start frame through a Region Proposal Network and then estimates the movements of proposals in next frame in a recurrent manner. The action proposals of different frames are linked to form the tubelet proposals. The RTR capitalizes on a multi-channel architecture, where in each channel, a tubelet proposal is fed into a CNN plus LSTM to recurrently recognize action in the tubelet. We conduct extensive experiments on four benchmark datasets and demonstrate superior results over state-of-the-art methods. More remarkably, we obtain mAP of 98.6 , 81.3 , 77.9 and 22.3 with gains of 2.9 , 4.3 , 0.7 and 3.9 over the best competitors on UCF-Sports, J-HMDB, UCF-101 and AVA, respectively.",
"Existing action detection algorithms usually generate action proposals through an extensive search over the video at multiple temporal scales, which brings about huge computational overhead and deviates from the human perception procedure. We argue that the process of detecting actions should be naturally one of observation and refinement: observe the current window and refine the span of attended window to cover true action regions. In this paper, we propose an active action proposal model that learns to find actions through continuously adjusting the temporal bounds in a self-adaptive way. The whole process can be deemed as an agent, which is firstly placed at a position in the video at random, adopts a sequence of transformations on the current attended region to discover actions according to a learned policy. We utilize reinforcement learning, especially the Deep Q-learning algorithm to learn the agent's decision policy. In addition, we use temporal pooling operation to extract more effective feature representation for the long temporal window, and design a regression network to adjust the position offsets between predicted results and the ground truth. Experiment results on THUMOS 2014 validate the effectiveness of the proposed approach, which can achieve competitive performance with current action detection algorithms via much fewer proposals.",
"Detecting actions in untrimmed videos is an important yet challenging task. In this paper, we present the structured segment network (SSN), a novel framework which models the temporal structure of each action instance via a structured temporal pyramid. On top of the pyramid, we further introduce a decomposed discriminative model comprising two classifiers, respectively for classifying actions and determining completeness. This allows the framework to effectively distinguish positive proposals from background or incomplete ones, thus leading to both accurate recognition and localization. These components are integrated into a unified network that can be efficiently trained in an end-to-end fashion. Additionally, a simple yet effective temporal action proposal scheme, dubbed temporal actionness grouping (TAG) is devised to generate high quality action proposals. On two challenging benchmarks, THUMOS14 and ActivityNet, our method remarkably outperforms previous state-of-the-art methods, demonstrating superior accuracy and strong adaptivity in handling actions with various temporal structures.",
"Detecting actions in untrimmed videos is an important yet challenging task. In this paper, we present the structured segment network (SSN), a novel framework which models the temporal structure of each action instance via a structured temporal pyramid. On top of the pyramid, we further introduce a decomposed discriminative model comprising two classifiers, respectively for classifying actions and determining completeness. This allows the framework to effectively distinguish positive proposals from background or incomplete ones, thus leading to both accurate recognition and localization. These components are integrated into a unified network that can be efficiently trained in an end-to-end fashion. Additionally, a simple yet effective temporal action proposal scheme, dubbed temporal actionness grouping (TAG) is devised to generate high quality action proposals. On two challenging benchmarks, THUMOS14 and ActivityNet, our method remarkably outperforms previous state-of-the-art methods, demonstrating superior accuracy and strong adaptivity in handling actions with various temporal structures.",
"In this work, we propose an approach to the spatiotemporal localisation (detection) and classification of multiple concurrent actions within temporally untrimmed videos. Our framework is composed of three stages. In stage 1, appearance and motion detection networks are employed to localise and score actions from colour images and optical flow. In stage 2, the appearance network detections are boosted by combining them with the motion detection scores, in proportion to their respective spatial overlap. In stage 3, sequences of detection boxes most likely to be associated with a single action instance, called action tubes, are constructed by solving two energy maximisation problems via dynamic programming. While in the first pass, action paths spanning the whole video are built by linking detection boxes over time using their class-specific scores and their spatial overlap, in the second pass, temporal trimming is performed by ensuring label consistency for all constituting detection boxes. We demonstrate the performance of our algorithm on the challenging UCF101, J-HMDB-21 and LIRIS-HARL datasets, achieving new state-of-the-art results across the board and significantly increasing detection speed at test time. We achieve a huge leap forward in action detection performance and report a 20 and 11 gain in mAP (mean average precision) on UCF-101 and J-HMDB-21 datasets respectively when compared to the state-of-the-art.",
"We propose a new task of unsupervised action detection by action matching. Given two long videos, the objective is to temporally detect all pairs of matching video segments. A pair of video segments are matched if they share the same human action. The task is category independent---it does not matter what action is being performed---and no supervision is used to discover such video segments. Unsupervised action detection by action matching allows us to align videos in a meaningful manner. As such, it can be used to discover new action categories or as an action proposal technique within, say, an action detection pipeline. Moreover, it is a useful pre-processing step for generating video highlights, e.g., from sports videos. We present an effective and efficient method for unsupervised action detection. We use an unsupervised temporal encoding method and exploit the temporal consistency in human actions to obtain candidate action segments. We evaluate our method on this challenging task using three activity recognition benchmarks, namely, the MPII Cooking activities dataset, the THUMOS15 action detection benchmark and a new dataset called the IKEA dataset. On the MPII Cooking dataset we detect action segments with a precision of 21.6 and recall of 11.7 over 946 long video pairs and over 5000 ground truth action segments. Similarly, on THUMOS dataset we obtain 18.4 precision and 25.1 recall over 5094 ground truth action segment pairs.",
"Dominant approaches to action detection can only provide sub-optimal solutions to the problem, as they rely on seeking frame-level detections, to later compose them into action tubes in a post-processing step. With this paper we radically depart from current practice, and take a first step towards the design and implementation of a deep network architecture able to classify and regress whole video subsets, so providing a truly optimal solution of the action detection problem. In this work, in particular, we propose a novel deep net framework able to regress and classify 3D region proposals spanning two consecutive video frames, whose core is an evolution of classical region proposal networks (RPNs). As such, our 3D-RPN net is able to effectively encode the temporal aspect of actions by purely exploiting appearance, as opposed to methods which heavily rely on expensive flow maps computed in a parallel stream. The proposed model is end-to-end trainable and can be jointly optimised for action localisation and classification using a single step of optimisation. At test time the network predicts 'micro-tubes' spanning two frames, which are linked up into complete action tubes via a new algorithm which exploits the temporal encoding learned by the network and cuts computation time by 50 . Promising results on the J-HMDB-21 and UCF-101 action detection datasets show that our model does indeed outperform the state-of-the-art when relying purely on appearance."
]
} |
1907.09706 | 2970623248 | Currently, the visually impaired rely on either a sighted human, guide dog, or white cane to safely navigate. However, the training of guide dogs is extremely expensive, and canes cannot provide essential information regarding the color of traffic lights and direction of crosswalks. In this paper, we propose a deep learning based solution that provides information regarding the traffic light mode and the position of the zebra crossing. Previous solutions that utilize machine learning only provide one piece of information and are mostly binary: only detecting red or green lights. The proposed convolutional neural network, LYTNet, is designed for comprehensiveness, accuracy, and computational efficiency. LYTNet delivers both of the two most important pieces of information for the visually impaired to cross the road. We provide five classes of pedestrian traffic lights rather than the commonly seen three or four, and a direction vector representing the midline of the zebra crossing that is converted from the 2D image plane to real-world positions. We created our own dataset of pedestrian traffic lights containing over 5000 photos taken at hundreds of intersections in Shanghai. The experiments carried out achieve a classification accuracy of 94 , average angle error of (6.35^ ), with a frame rate of 20 frames per second when testing the network on an iPhone 7 with additional post-processing steps. | Some industrialized countries have developed acoustic pedestrian traffic lights that produce sound when the light is green, and is used as a signal for the visually impaired to know when to cross the street @cite_14 @cite_17 @cite_3 . However, for less economically developed countries, crossing streets is still a problem for the blind, and acoustic pedestrian traffic lights are not ubiquitous even in developed nations @cite_14 . | {
"cite_N": [
"@cite_14",
"@cite_3",
"@cite_17"
],
"mid": [
"2115104908",
"201681701",
"2771992117"
],
"abstract": [
"Pedestrians' use of Motion Pictures Expert Group audio layer 3 players or mobile phones can pose the risk of being hit by motor vehicles. We present an approach for detecting a crash risk level using the computing power and the microphone of mobile devices that can be used to alert the user in advance of an approaching vehicle so as to avoid a crash. A single feature extractor classifier is not usually able to deal with the diversity of risky acoustic scenarios. In this paper, we address the problem of detection of vehicles approaching a pedestrian by a novel simple nonresource intensive acoustic method. The method uses a set of existing statistical tools to mine signal features. Audio features are adaptively thresholded for relevance and classified with a three-component heuristic. The resulting acoustic hazard detection system has a very low false-positive detection rate. The results of this study could help mobile device manufacturers to embed the presented features into future potable devices and contribute to road safety.",
"Urban intersections are the most dangerous parts of a blind or visually impaired person's travel. To address this problem, this paper describes the novel \"Crosswatch\" system, which uses computer vision to provide information about the location and orientation of crosswalks to a blind or visually impaired pedestrian holding a camera cell phone. A prototype of the system runs on an off-the-shelf Nokia camera phone in real time, which automatically takes a few images per second, uses the cell phone's built-in computer to analyze each image in a fraction of a second and sounds an audio tone when it detects a crosswalk. Tests with blind subjects demonstrate the feasibility of the system and its ability to provide useful crosswalk alignment information under real-world conditions.",
"In defect of intelligent assistant approaches, the visually impaired feel hard to cross the roads in urban environments. Aiming to tackle the problem, a real-time Pedestrian Crossing Lights (PCL) detection algorithm for the visually impaired is proposed in this paper. Different from previous works which utilize analytic image processing to detect the PCL in ideal scenarios, the proposed algorithm detects PCL using machine learning scheme in the challenging scenarios, where PCL have arbitrary sizes and locations in acquired image and suffer from the shake and movement of camera. In order to achieve the robustness and efficiency in those scenarios, the detection algorithm is designed to include three procedures: candidate extraction, candidate recognition and temporal-spatial analysis. A public dataset of PCL, which includes manually labeled ground truth data, is established for tuning parameters, training samples and evaluating the performance. The algorithm is implemented on a portable PC with color camera. The experiments carried out in various practical scenarios prove that the precision and recall of detection are both close to 100 , meanwhile the frame rate is up to 21 frames per second (FPS)."
]
} |
1907.09706 | 2970623248 | Currently, the visually impaired rely on either a sighted human, guide dog, or white cane to safely navigate. However, the training of guide dogs is extremely expensive, and canes cannot provide essential information regarding the color of traffic lights and direction of crosswalks. In this paper, we propose a deep learning based solution that provides information regarding the traffic light mode and the position of the zebra crossing. Previous solutions that utilize machine learning only provide one piece of information and are mostly binary: only detecting red or green lights. The proposed convolutional neural network, LYTNet, is designed for comprehensiveness, accuracy, and computational efficiency. LYTNet delivers both of the two most important pieces of information for the visually impaired to cross the road. We provide five classes of pedestrian traffic lights rather than the commonly seen three or four, and a direction vector representing the midline of the zebra crossing that is converted from the 2D image plane to real-world positions. We created our own dataset of pedestrian traffic lights containing over 5000 photos taken at hundreds of intersections in Shanghai. The experiments carried out achieve a classification accuracy of 94 , average angle error of (6.35^ ), with a frame rate of 20 frames per second when testing the network on an iPhone 7 with additional post-processing steps. | The task of detecting traffic light for autonomous driving has been explored by many and has developed over the years @cite_2 @cite_15 @cite_1 @cite_7 . @cite_8 created a model that is able to detect traffic lights as small as @math pixels and with relatively high accuracy. Though most models for traffic lights have a high precision and recall rate of nearly 100 @cite_13 were one of the first to develop an algorithm to detect pedestrian traffic lights and the length of the zebra-crossing. Others such as and @cite_14 @cite_16 both developed an analytic image processing algorithm, which undergoes candidate extraction, candidate recognition, and candidate classification. @cite_3 proposed a more robust real-time pedestrian traffic lights detection algorithm, which gets rid of the analytic image processing method and uses candidate extraction and a concise machine learning scheme. | {
"cite_N": [
"@cite_14",
"@cite_7",
"@cite_8",
"@cite_1",
"@cite_3",
"@cite_2",
"@cite_15",
"@cite_16",
"@cite_13"
],
"mid": [
"2963315052",
"2184993296",
"1968974896",
"2737202447",
"2797363357",
"2964332990",
"2757356831",
"2771992117",
"1526734559"
],
"abstract": [
"In this paper, we consider the problem of pedestrian detection in natural scenes. Intuitively, instances of pedestrians with different spatial scales may exhibit dramatically different features. Thus, large variance in instance scales, which results in undesirable large intracategory variance in features, may severely hurt the performance of modern object instance detection methods. We argue that this issue can be substantially alleviated by the divide-and-conquer philosophy. Taking pedestrian detection as an example, we illustrate how we can leverage this philosophy to develop a Scale-Aware Fast R-CNN (SAF R-CNN) framework. The model introduces multiple built-in subnetworks which detect pedestrians with scales from disjoint ranges. Outputs from all of the subnetworks are then adaptively combined to generate the final detection results that are shown to be robust to large variance in instance scales, via a gate function defined over the sizes of object proposals. Extensive evaluations on several challenging pedestrian detection datasets well demonstrate the effectiveness of the proposed SAF R-CNN. Particularly, our method achieves state-of-the-art performance on Caltech [P. Dollar, C. Wojek, B. Schiele, and P. Perona, “Pedestrian detection: An evaluation of the state of the art,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 34, no. 4, pp. 743–761, Apr. 2012], and obtains competitive results on INRIA [N. Dalal and B. Triggs, “Histograms of oriented gradients for human detection,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. , 2005, pp. 886–893], ETH [A. Ess, B. Leibe, and L. V. Gool, “Depth and appearance for mobile scene analysis,” in Proc. Int. Conf. Comput. Vis ., 2007, pp. 1–8], and KITTI [A. Geiger, P. Lenz, and R. Urtasun, “Are we ready for autonomous driving? The KITTI vision benchmark suite,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit ., 2012, pp. 3354–3361].",
"TL-recognizer detects traffic lights from a mobile device camera.Robust method for unsupervised image acquisition and segmentation.Robust solution: traffic lights are clearly visible in different light conditions.Solution is reliable: precision 1 and recall 0.8 in different light conditions.Solution is efficient: computation time 100źms on a Nexus 5. Independent mobility involves a number of challenges for people with visual impairment or blindness. In particular, in many countries the majority of traffic lights are still not equipped with acoustic signals. Recognizing traffic lights through the analysis of images acquired by a mobile device camera is a viable solution already experimented in scientific literature. However, there is a major issue: the recognition techniques should be robust under different illumination conditions.This contribution addresses the above problem with an effective solution: besides image processing and recognition, it proposes a robust setup for image capture that makes it possible to acquire clearly visible traffic light images regardless of daylight variability due to time and weather. The proposed recognition technique that adopts this approach is reliable (full precision and high recall), robust (works in different illumination conditions) and efficient (it can run several times a second on commercial smartphones). The experimental evaluation conducted with visual impaired subjects shows that the technique is also practical in supporting road crossing.",
"The recognition and tracking of traffic lights for intelligent vehicles based on a vehicle-mounted camera are studied in this paper. The candidate region of the traffic light is extracted using the threshold segmentation method and the morphological operation. Then, the recognition algorithm of the traffic light based on machine learning is employed. To avoid false negatives and tracking loss, the target tracking algorithm CAMSHIFT (Continuously Adaptive Mean Shift), which uses the color histogram as the target model, is adopted. In addition to traffic signal pre-processing and the recognition method of learning, the initialization problem of the search window of CAMSHIFT algorithm is resolved. Moreover, the window setting method is used to shorten the processing time of the global HSV color space conversion. The real vehicle experiments validate the performance of the presented approach.",
"Reliable traffic light detection and classification is crucial for automated driving in urban environments. Currently, there are no systems that can reliably perceive traffic lights in real-time, without map-based information, and in sufficient distances needed for smooth urban driving. We propose a complete system consisting of a traffic light detector, tracker, and classifier based on deep learning, stereo vision, and vehicle odometry which perceives traffic lights in real-time. Within the scope of this work, we present three major contributions. The first is an accurately labeled traffic light dataset of 5000 images for training and a video sequence of 8334 frames for evaluation. The dataset is published as the Bosch Small Traffic Lights Dataset and uses our results as baseline. It is currently the largest publicly available labeled traffic light dataset and includes labels down to the size of only 1 pixel in width. The second contribution is a traffic light detector which runs at 10 frames per second on 1280×720 images. When selecting the confidence threshold that yields equal error rate, we are able to detect traffic lights as small as 4 pixels in width. The third contribution is a traffic light tracker which uses stereo vision and vehicle odometry to compute the motion estimate of traffic lights and a neural network to correct the aforementioned motion estimate.",
"Traffic lights detection and recognition research has grown every year. Time is coming when autonomous vehicle can navigate in urban roads and streets and intelligent systems aboard those cars would have to recognize traffic lights in real time. This article proposes a traffic light recognition (TLR) device prototype using a smartphone as camera and processing unit that can be used as a driver assistance. A TLR device has to be able to visualize the traffic scene from inside of a vehicle, generate stable images, and be protected from adverse conditions. To validate this layout prototype, a dataset was built and used to test an algorithm that uses an adaptive background suppression filter (AdaBSF) and Support Vector Machines (SVMs) to detect traffic lights. The application of AdaBSF and subsequent classification with SVM to the dataset achieved 100 precision rate and recall of 65 . Road testing shows that the TLR device prototype meets the requirements to be used as a driver assistance device.",
"In this paper, we propose a unified end-to-end trainable multi-task network that jointly handles lane and road marking detection and recognition that is guided by a vanishing point under adverse weather conditions. We tackle rainy and low illumination conditions, which have not been extensively studied until now due to clear challenges. For example, images taken under rainy days are subject to low illumination, while wet roads cause light reflection and distort the appearance of lane and road markings. At night, color distortion occurs under limited illumination. As a result, no benchmark dataset exists and only a few developed algorithms work under poor weather conditions. To address this shortcoming, we build up a lane and road marking benchmark which consists of about 20,000 images with 17 lane and road marking classes under four different scenarios: no rain, rain, heavy rain, and night. We train and evaluate several versions of the proposed multi-task network and validate the importance of each task. The resulting approach, VPGNet, can detect and classify lanes and road markings, and predict a vanishing point with a single forward pass. Experimental results show that our approach achieves high accuracy and robustness under various conditions in realtime (20 fps). The benchmark and the VPGNet model will be publicly available",
"In this paper, we propose a unified end-to-end trainable multi-task network that jointly handles lane and road marking detection and recognition that is guided by a vanishing point under adverse weather conditions. We tackle rainy and low illumination conditions, which have not been extensively studied until now due to clear challenges. For example, images taken under rainy days are subject to low illumination, while wet roads cause light reflection and distort the appearance of lane and road markings. At night, color distortion occurs under limited illumination. As a result, no benchmark dataset exists and only a few developed algorithms work under poor weather conditions. To address this shortcoming, we build up a lane and road marking benchmark which consists of about 20,000 images with 17 lane and road marking classes under four different scenarios: no rain, rain, heavy rain, and night. We train and evaluate several versions of the proposed multi-task network and validate the importance of each task. The resulting approach, VPGNet, can detect and classify lanes and road markings, and predict a vanishing point with a single forward pass. Experimental results show that our approach achieves high accuracy and robustness under various conditions in real-time (20 fps). The benchmark and the VPGNet model will be publicly available.",
"In defect of intelligent assistant approaches, the visually impaired feel hard to cross the roads in urban environments. Aiming to tackle the problem, a real-time Pedestrian Crossing Lights (PCL) detection algorithm for the visually impaired is proposed in this paper. Different from previous works which utilize analytic image processing to detect the PCL in ideal scenarios, the proposed algorithm detects PCL using machine learning scheme in the challenging scenarios, where PCL have arbitrary sizes and locations in acquired image and suffer from the shake and movement of camera. In order to achieve the robustness and efficiency in those scenarios, the detection algorithm is designed to include three procedures: candidate extraction, candidate recognition and temporal-spatial analysis. A public dataset of PCL, which includes manually labeled ground truth data, is established for tuning parameters, training samples and evaluating the performance. The algorithm is implemented on a portable PC with color camera. The experiments carried out in various practical scenarios prove that the precision and recall of detection are both close to 100 , meanwhile the frame rate is up to 21 frames per second (FPS).",
"Pedestrian detection is of crucial importance to autonomous driving applications. Methods based on deep learning have shown significant improvements in accuracy, which makes them particularly suitable for applications, such as pedestrian detection, where reducing the miss rate is very important. Although they are accurate, their runtime has been at best in seconds per image, which makes them not practical for onboard applications. We present a Large-Field-Of-View (LFOV) deep network for pedestrian detection, that can achieve high accuracy and is designed to make deep networks work faster for detection problems. The idea of the proposed Large-Field-of-View deep network is to learn to make classification decisions simultaneously and accurately at multiple locations. The LFOV network processes larger image areas at much faster speeds than typical deep networks have been able to, and can intrinsically reuse computations. Our pedestrian detection solution, which is a combination of a LFOV network and a standard deep network, works at 280 ms per image on GPU and achieves 35.85 average miss rate on the Caltech Pedestrian Detection Benchmark."
]
} |
1907.09706 | 2970623248 | Currently, the visually impaired rely on either a sighted human, guide dog, or white cane to safely navigate. However, the training of guide dogs is extremely expensive, and canes cannot provide essential information regarding the color of traffic lights and direction of crosswalks. In this paper, we propose a deep learning based solution that provides information regarding the traffic light mode and the position of the zebra crossing. Previous solutions that utilize machine learning only provide one piece of information and are mostly binary: only detecting red or green lights. The proposed convolutional neural network, LYTNet, is designed for comprehensiveness, accuracy, and computational efficiency. LYTNet delivers both of the two most important pieces of information for the visually impaired to cross the road. We provide five classes of pedestrian traffic lights rather than the commonly seen three or four, and a direction vector representing the midline of the zebra crossing that is converted from the 2D image plane to real-world positions. We created our own dataset of pedestrian traffic lights containing over 5000 photos taken at hundreds of intersections in Shanghai. The experiments carried out achieve a classification accuracy of 94 , average angle error of (6.35^ ), with a frame rate of 20 frames per second when testing the network on an iPhone 7 with additional post-processing steps. | A limitation that many attempts faced was the speed of hardware. Thus, @cite_5 created an algorithm specifically for mobile devices with an accelerator to detect pedestrian traffic lights in real time. @cite_4 incorporated external servers to remove the limitation of hardware and provide more accurate information. Though the external servers are able to run deeper models than phones, it requires fast and stable internet connection at all times. Moreover, the advancement of efficient neural networks such as MobileNet v2 enable a deep-learning approach to be implemented on a mobile device @cite_11 . | {
"cite_N": [
"@cite_5",
"@cite_4",
"@cite_11"
],
"mid": [
"2265127172",
"2963315052",
"2944779197"
],
"abstract": [
"We present a new real-time approach to object detection that exploits the efficiency of cascade classifiers with the accuracy of deep neural networks. Deep networks have been shown to excel at classification tasks, and their ability to operate on raw pixel input without the need to design special features is very appealing. However, deep nets are notoriously slow at inference time. In this paper, we propose an approach that cascades deep nets and fast features, that is both very fast and very accurate. We apply it to the challenging task of pedestrian detection. Our algorithm runs in real-time at 15 frames per second. The resulting approach achieves a 26.2 average miss rate on the Caltech Pedestrian detection benchmark, which is competitive with the very best reported results. It is the first work we are aware of that achieves very high accuracy while running in real-time.",
"In this paper, we consider the problem of pedestrian detection in natural scenes. Intuitively, instances of pedestrians with different spatial scales may exhibit dramatically different features. Thus, large variance in instance scales, which results in undesirable large intracategory variance in features, may severely hurt the performance of modern object instance detection methods. We argue that this issue can be substantially alleviated by the divide-and-conquer philosophy. Taking pedestrian detection as an example, we illustrate how we can leverage this philosophy to develop a Scale-Aware Fast R-CNN (SAF R-CNN) framework. The model introduces multiple built-in subnetworks which detect pedestrians with scales from disjoint ranges. Outputs from all of the subnetworks are then adaptively combined to generate the final detection results that are shown to be robust to large variance in instance scales, via a gate function defined over the sizes of object proposals. Extensive evaluations on several challenging pedestrian detection datasets well demonstrate the effectiveness of the proposed SAF R-CNN. Particularly, our method achieves state-of-the-art performance on Caltech [P. Dollar, C. Wojek, B. Schiele, and P. Perona, “Pedestrian detection: An evaluation of the state of the art,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 34, no. 4, pp. 743–761, Apr. 2012], and obtains competitive results on INRIA [N. Dalal and B. Triggs, “Histograms of oriented gradients for human detection,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. , 2005, pp. 886–893], ETH [A. Ess, B. Leibe, and L. V. Gool, “Depth and appearance for mobile scene analysis,” in Proc. Int. Conf. Comput. Vis ., 2007, pp. 1–8], and KITTI [A. Geiger, P. Lenz, and R. Urtasun, “Are we ready for autonomous driving? The KITTI vision benchmark suite,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit ., 2012, pp. 3354–3361].",
"We present the next generation of MobileNets based on a combination of complementary search techniques as well as a novel architecture design. MobileNetV3 is tuned to mobile phone CPUs through a combination of hardware-aware network architecture search (NAS) complemented by the NetAdapt algorithm and then subsequently improved through novel architecture advances. This paper starts the exploration of how automated search algorithms and network design can work together to harness complementary approaches improving the overall state of the art. Through this process we create two new MobileNet models for release: MobileNetV3-Large and MobileNetV3-Small which are targeted for high and low resource use cases. These models are then adapted and applied to the tasks of object detection and semantic segmentation. For the task of semantic segmentation (or any dense pixel prediction), we propose a new efficient segmentation decoder Lite Reduced Atrous Spatial Pyramid Pooling (LR-ASPP). We achieve new state of the art results for mobile classification, detection and segmentation. MobileNetV3-Large is 3.2 more accurate on ImageNet classification while reducing latency by 15 compared to MobileNetV2. MobileNetV3-Small is 4.6 more accurate while reducing latency by 5 compared to MobileNetV2. MobileNetV3-Large detection is 25 faster at roughly the same accuracy as MobileNetV2 on COCO detection. MobileNetV3-Large LR-ASPP is 30 faster than MobileNetV2 R-ASPP at similar accuracy for Cityscapes segmentation."
]
} |
1907.09706 | 2970623248 | Currently, the visually impaired rely on either a sighted human, guide dog, or white cane to safely navigate. However, the training of guide dogs is extremely expensive, and canes cannot provide essential information regarding the color of traffic lights and direction of crosswalks. In this paper, we propose a deep learning based solution that provides information regarding the traffic light mode and the position of the zebra crossing. Previous solutions that utilize machine learning only provide one piece of information and are mostly binary: only detecting red or green lights. The proposed convolutional neural network, LYTNet, is designed for comprehensiveness, accuracy, and computational efficiency. LYTNet delivers both of the two most important pieces of information for the visually impaired to cross the road. We provide five classes of pedestrian traffic lights rather than the commonly seen three or four, and a direction vector representing the midline of the zebra crossing that is converted from the 2D image plane to real-world positions. We created our own dataset of pedestrian traffic lights containing over 5000 photos taken at hundreds of intersections in Shanghai. The experiments carried out achieve a classification accuracy of 94 , average angle error of (6.35^ ), with a frame rate of 20 frames per second when testing the network on an iPhone 7 with additional post-processing steps. | Direction is another factor to consider when helping the visually impaired cross the street. Though the visually impaired can have a good sense of the general direction to cross the road in familiar environments, relying on one's memory has its limitations @cite_6 . Therefore, solutions to provide specific direction have also been devised. Other than detecting the color of pedestrian traffic lights, @cite_6 also created an algorithm for detecting zebra crossings. The system obtains information of how much of the zebra-crossing is visible to help the visually impaired know whether or not they are generally facing in the correct direction, but it does not provide the specific location of the zebra crossing. , , and Banich @cite_9 @cite_10 @cite_0 also use deep learning neural network within computer vision to detect zebra crossings to help the visually impaired cross streets. However, no deep learning method is able to output both traffic light and zebra crossing information simultaneously. | {
"cite_N": [
"@cite_0",
"@cite_9",
"@cite_10",
"@cite_6"
],
"mid": [
"2888662204",
"2100894161",
"2771992117",
"1519128923"
],
"abstract": [
"For mobile robots navigating on sidewalks, it is essential to be able to safely cross street intersections. Most existing approaches rely on the recognition of the traffic light signal to make an informed crossing decision. Although these approaches have been crucial enablers for urban navigation, the capabilities of robots employing such approaches are still limited to navigating only on streets containing signalized intersections. In this paper, we address this challenge and propose a multimodal convolutional neural network framework to predict the safety of a street intersection for crossing. Our architecture consists of two subnetworks; an interaction-aware trajectory estimation stream IA-TCNN, that predicts the future states of all observed traffic participants in the scene, and a traffic light recognition stream AtteNet. Our IA-TCNN utilizes dilated causal convolutions to model the behavior of the observable dynamic agents in the scene without explicitly assigning priorities to the interactions among them. While AtteNet utilizes Squeeze-Excitation blocks to learn a content-aware mechanism for selecting the relevant features from the data, thereby improving the noise robustness. Learned representations from the traffic light recognition stream are fused with the estimated trajectories from the motion prediction stream to learn the crossing decision. Furthermore, we extend our previously introduced Freiburg Street Crossing dataset with sequences captured at different types of intersections, demonstrating complex interactions among the traffic participants. Extensive experimental evaluations on public benchmark datasets and our proposed dataset demonstrate that our network achieves state-of-the-art performance for each of the subtasks, as well as for the crossing safety prediction.",
"In smart-cities, computer vision has the potential to dramatically improve the quality of life of people suffering of visual impairments. In this field, we have been working on a wearable mobility aid aimed at detecting in real-time obstacles in front of a visually impaired. Our approach relies on a custom RGBD camera, with FPGA on-board processing, worn as traditional eyeglasses and effective point-cloud processing implemented on a compact and lightweight embedded computer. This latter device also provides feedback to the user by means of an haptic interface as well as audio messages. In this paper we address crosswalk recognition that, as pointed out by several visually impaired users involved in the evaluation of our system, is a crucial requirement in the design of an effective mobility aid. Specifically, we propose a reliable methodology to detect and categorize crosswalks by leveraging on point-cloud processing and deep-learning techniques. The experimental results reported, on 10000+ frames, confirm that the proposed approach is invariant to head camera pose and extremely effective even when dealing with large occlusions typically found in urban environments.",
"In defect of intelligent assistant approaches, the visually impaired feel hard to cross the roads in urban environments. Aiming to tackle the problem, a real-time Pedestrian Crossing Lights (PCL) detection algorithm for the visually impaired is proposed in this paper. Different from previous works which utilize analytic image processing to detect the PCL in ideal scenarios, the proposed algorithm detects PCL using machine learning scheme in the challenging scenarios, where PCL have arbitrary sizes and locations in acquired image and suffer from the shake and movement of camera. In order to achieve the robustness and efficiency in those scenarios, the detection algorithm is designed to include three procedures: candidate extraction, candidate recognition and temporal-spatial analysis. A public dataset of PCL, which includes manually labeled ground truth data, is established for tuning parameters, training samples and evaluating the performance. The algorithm is implemented on a portable PC with color camera. The experiments carried out in various practical scenarios prove that the precision and recall of detection are both close to 100 , meanwhile the frame rate is up to 21 frames per second (FPS).",
"Despite recent significant advances, pedestrian detection continues to be an extremely challenging problem in real scenarios. In order to develop a detector that successfully operates under these conditions, it becomes critical to leverage upon multiple cues, multiple imaging modalities and a strong multi-view classifier that accounts for different pedestrian views and poses. In this paper we provide an extensive evaluation that gives insight into how each of these aspects (multi-cue, multi-modality and strong multi-view classifier) affect performance both individually and when integrated together. In the multi-modality component we explore the fusion of RGB and depth maps obtained by high-definition LIDAR, a type of modality that is only recently starting to receive attention. As our analysis reveals, although all the aforementioned aspects significantly help in improving the performance, the fusion of visible spectrum and depth information allows to boost the accuracy by a much larger margin. The resulting detector not only ranks among the top best performers in the challenging KITTI benchmark, but it is built upon very simple blocks that are easy to implement and computationally efficient. These simple blocks can be easily replaced with more sophisticated ones recently proposed, such as the use of convolutional neural networks for feature representation, to further improve the accuracy."
]
} |
1907.09642 | 2963590785 | Image smoothing is a fundamental procedure in applications of both computer vision and graphics. The required smoothing properties can be different or even contradictive among different tasks. Nevertheless, the inherent smoothing nature of one smoothing operator is usually fixed and thus cannot meet the various requirements of different applications. In this paper, a non-convex non-smooth optimization framework is proposed to achieve diverse smoothing natures where even contradictive smoothing behaviors can be achieved. To this end, we first introduce the truncated Huber penalty function which has seldom been used in image smoothing. A robust framework is then proposed. When combined with the strong flexibility of the truncated Huber penalty function, our framework is capable of a range of applications and can outperform the state-of-the-art approaches in several tasks. In addition, an efficient numerical solution is provided and its convergence is theoretically guaranteed even the optimization framework is non-convex and non-smooth. The effectiveness and superior performance of our approach are validated through comprehensive experimental results in a range of applications. | In terms of structure-preserving smoothing, Zhang et al. @cite_28 proposed to smooth structures of different scales with a rolling guidance filter (RGF). Cho et al. @cite_27 modified the original BLF with local patch-based analysis of texture features and obtained a bilateral texture filter (BTF) for image texture removal. Karacan et al. @cite_7 proposed to smooth image textures by making use of region covariances that captured local structure and textural information. Xu et al. @cite_8 adopted the relative total variation (RTV) as a prior to regularize the texture smoothing procedure. Chen et al. @cite_10 proved that the TV- @math model @cite_10 @cite_43 could smooth images in a scale-aware manner, and it is thus ideal for structure-preserving smoothing such as image texture removal @cite_1 @cite_33 . | {
"cite_N": [
"@cite_33",
"@cite_7",
"@cite_8",
"@cite_28",
"@cite_1",
"@cite_43",
"@cite_27",
"@cite_10"
],
"mid": [
"2109075629",
"2057203222",
"1909952827",
"1482080565",
"2892998444",
"1974438823",
"832925222",
"2231495490"
],
"abstract": [
"This paper presents a novel structure-preserving image decomposition operator called bilateral texture filter. As a simple modification of the original bilateral filter [Tomasi and Manduchi 1998], it performs local patch-based analysis of texture features and incorporates its results into the range filter kernel. The central idea to ensure proper texture structure separation is based on patch shift that captures the texture information from the most representative texture patch clear of prominent structure edges. Our method outperforms the original bilateral filter in removing texture while preserving main image structures, at the cost of some added computation. It inherits well-known advantages of the bilateral filter, such as simplicity, local nature, ease of implementation, scalability, and adaptability to other application scenarios.",
"Recent years have witnessed the emergence of new image smoothing techniques which have provided new insights and raised new questions about the nature of this well-studied problem. Specifically, these models separate a given image into its structure and texture layers by utilizing non-gradient based definitions for edges or special measures that distinguish edges from oscillations. In this study, we propose an alternative yet simple image smoothing approach which depends on covariance matrices of simple image features, aka the region covariances. The use of second order statistics as a patch descriptor allows us to implicitly capture local structure and texture information and makes our approach particularly effective for structure extraction from texture. Our experimental results have shown that the proposed approach leads to better image decompositions as compared to the state-of-the-art methods and preserves prominent edges and shading well. Moreover, we also demonstrate the applicability of our approach on some image editing and manipulation tasks such as image abstraction, texture and detail enhancement, image composition, inverse halftoning and seam carving.",
"Research in texture recognition often concentrates on the problem of material recognition in uncluttered conditions, an assumption rarely met by applications. In this work we conduct a first study of material and describable texture attributes recognition in clutter, using a new dataset derived from the OpenSurface texture repository. Motivated by the challenge posed by this problem, we propose a new texture descriptor, FV-CNN, obtained by Fisher Vector pooling of a Convolutional Neural Network (CNN) filter bank. FV-CNN substantially improves the state-of-the-art in texture, material and scene recognition. Our approach achieves 79.8 accuracy on Flickr material dataset and 81 accuracy on MIT indoor scenes, providing absolute gains of more than 10 over existing approaches. FV-CNN easily transfers across domains without requiring feature adaptation as for methods that build on the fully-connected layers of CNNs. Furthermore, FV-CNN can seamlessly incorporate multi-scale information and describe regions of arbitrary shapes and sizes. Our approach is particularly suited at localizing “stuff” categories and obtains state-of-the-art results on MSRC segmentation dataset, as well as promising results on recognizing materials and surface attributes in clutter on the OpenSurfaces dataset.",
"Research in texture recognition often concentrates on the problem of material recognition in uncluttered conditions, an assumption rarely met by applications. In this work we conduct a first study of material and describable texture at- tributes recognition in clutter, using a new dataset derived from the OpenSurface texture repository. Motivated by the challenge posed by this problem, we propose a new texture descriptor, D-CNN, obtained by Fisher Vector pooling of a Convolutional Neural Network (CNN) filter bank. D-CNN substantially improves the state-of-the-art in texture, mate- rial and scene recognition. Our approach achieves 82.3 accuracy on Flickr material dataset and 81.1 accuracy on MIT indoor scenes, providing absolute gains of more than 10 over existing approaches. D-CNN easily trans- fers across domains without requiring feature adaptation as for methods that build on the fully-connected layers of CNNs. Furthermore, D-CNN can seamlessly incorporate multi-scale information and describe regions of arbitrary shapes and sizes. Our approach is particularly suited at lo- calizing stuff categories and obtains state-of-the-art re- sults on MSRC segmentation dataset, as well as promising results on recognizing materials and surface attributes in clutter on the OpenSurfaces dataset.",
"The performance of single image super-resolution has achieved significant improvement by utilizing deep convolutional neural networks (CNNs). The features in deep CNN contain different types of information which make different contributions to image reconstruction. However, most CNN-based models lack discriminative ability for different types of information and deal with them equally, which results in the representational capacity of the models being limited. On the other hand, as the depth of neural networks grows, the long-term information coming from preceding layers is easy to be weaken or lost in late layers, which is adverse to super-resolving image. To capture more informative features and maintain long-term information for image super-resolution, we propose a channel-wise and spatial feature modulation (CSFM) network in which a sequence of feature-modulation memory (FMM) modules is cascaded with a densely connected structure to transform low-resolution features to high informative features. In each FMM module, we construct a set of channel-wise and spatial attention residual (CSAR) blocks and stack them in a chain structure to dynamically modulate multi-level features in a global-and-local manner. This feature modulation strategy enables the high contribution information to be enhanced and the redundant information to be suppressed. Meanwhile, for long-term information persistence, a gated fusion (GF) node is attached at the end of the FMM module to adaptively fuse hierarchical features and distill more effective information via the dense skip connections and the gating mechanism. Extensive quantitative and qualitative evaluations on benchmark datasets illustrate the superiority of our proposed method over the state-of-the-art methods.",
"We present an extension of the BM3D filter to volumetric data. The proposed algorithm, BM4D, implements the grouping and collaborative filtering paradigm, where mutually similar d -dimensional patches are stacked together in a (d+1) -dimensional array and jointly filtered in transform domain. While in BM3D the basic data patches are blocks of pixels, in BM4D we utilize cubes of voxels, which are stacked into a 4-D “group.” The 4-D transform applied on the group simultaneously exploits the local correlation present among voxels in each cube and the nonlocal correlation between the corresponding voxels of different cubes. Thus, the spectrum of the group is highly sparse, leading to very effective separation of signal and noise through coefficient shrinkage. After inverse transformation, we obtain estimates of each grouped cube, which are then adaptively aggregated at their original locations. We evaluate the algorithm on denoising of volumetric data corrupted by Gaussian and Rician noise, as well as on reconstruction of volumetric phantom data with non-zero phase from noisy and incomplete Fourier-domain (k-space) measurements. Experimental results demonstrate the state-of-the-art denoising performance of BM4D, and its effectiveness when exploited as a regularizer in volumetric data reconstruction.",
"Accurate recovery of 3D geometrical surfaces from calibrated 2D multi-view images is a fundamental yet active research area in computer vision. Despite the steady progress in multi-view stereo (MVS) reconstruction, many existing methods are still limited in recovering fine-scale details and sharp features while suppressing noises, and may fail in reconstructing regions with less textures. To address these limitations, this paper presents a detail-preserving and content-aware variational (DCV) MVS method, which reconstructs the 3D surface by alternating between reprojection error minimization and mesh denoising. In reprojection error minimization, we propose a novel inter-image similarity measure, which is effective to preserve fine-scale details of the reconstructed surface and builds a connection between guided image filtering and image registration. In mesh denoising, we propose a content-aware @math -minimization algorithm by adaptively estimating the @math value and regularization parameters. Compared with conventional isotropic mesh smoothing approaches, the proposed method is much more promising in suppressing noise while preserving sharp features. Experimental results on benchmark data sets demonstrate that our DCV method is capable of recovering more surface details, and obtains cleaner and more accurate reconstructions than the state-of-the-art methods. In particular, our method achieves the best results among all published methods on the Middlebury dino ring and dino sparse data sets in terms of both completeness and accuracy.",
"Images contain many levels of important structures and edges. Compared to masses of research to make filters edge preserving, finding scale-aware local operations was seldom addressed in a practical way, albeit similarly vital in image processing and computer vision. We propose a new framework to filter images with the complete control of detail smoothing under a scale measure. It is based on a rolling guidance implemented in an iterative manner that converges quickly. Our method is simple in implementation, easy to understand, fully extensible to accommodate various data operations, and fast to produce results. Our implementation achieves realtime performance and produces artifact-free results in separating different scale structures. This filter also introduces several inspiring properties different from previous edge-preserving ones."
]
} |
1907.09478 | 2964056701 | Digital histology images are amenable to the application of convolutional neural network (CNN) for analysis due to the sheer size of pixel data present in them. CNNs are generally used for representation learning from small image patches (e.g. 224x224) extracted from digital histology images due to computational and memory constraints. However, this approach does not incorporate high-resolution contextual information in histology images. We propose a novel way to incorporate larger context by a context-aware neural network based on images with a dimension of 1,792x1,792 pixels. The proposed framework first encodes the local representation of a histology image into high dimensional features then aggregates the features by considering their spatial organization to make a final prediction. The proposed method is evaluated for colorectal cancer grading and breast cancer classification. A comprehensive analysis of some variants of the proposed method is presented. Our method outperformed the traditional patch-based approaches, problem-specific methods, and existing context-based methods quantitatively by a margin of 3.61 . Code and dataset related information is available at this link: this https URL | In literature, various different approaches have been presented to incorporate the contextual information for the classification of histology images. Some researchers @cite_28 @cite_29 @cite_7 used image down-sampling, a common practice followed in natural image classification, to capture the context from larger histology image. However, this approach is not suitable for problems where cell information is as important as the context. Adaptive patch sampling @cite_18 and discriminative patch selection @cite_21 from histology images is another way to integrate the sparse context. These methods are not capable of capturing small regions of interest at high resolution e.g. tumor cells and their local contextual arrangement. Some methods @cite_6 @cite_23 @cite_9 @cite_17 leverage the multi-resolution nature of histology images and use multi-resolution based classifiers to capture context. These multi-resolution approaches only consider a small part of an image at high resolution and the remaining part at low resolutions to make a prediction. Therefore, these approaches lack the contextual information of cellular architecture at high resolution in a histology image. | {
"cite_N": [
"@cite_18",
"@cite_7",
"@cite_28",
"@cite_29",
"@cite_21",
"@cite_9",
"@cite_6",
"@cite_23",
"@cite_17"
],
"mid": [
"2036924016",
"1975020933",
"2098140880",
"2612377680",
"2890554424",
"2129259959",
"2897601127",
"2173930701",
"2949362008"
],
"abstract": [
"Automatic analysis of histopathological images has been widely utilized leveraging computational image-processing methods and modern machine learning techniques. Both computer-aided diagnosis (CAD) and content-based image-retrieval (CBIR) systems have been successfully developed for diagnosis, disease detection, and decision support in this area. Recently, with the ever-increasing amount of annotated medical data, large-scale and data-driven methods have emerged to offer a promise of bridging the semantic gap between images and diagnostic information. In this paper, we focus on developing scalable image-retrieval techniques to cope intelligently with massive histopathological images. Specifically, we present a supervised kernel hashing technique which leverages a small amount of supervised information in learning to compress a 10 @math 000-dimensional image feature vector into only tens of binary bits with the informative signatures preserved. These binary codes are then indexed into a hash table that enables real-time retrieval of images in a large database. Critically, the supervised information is employed to bridge the semantic gap between low-level image features and high-level diagnostic information. We build a scalable image-retrieval framework based on the supervised hashing technique and validate its performance on several thousand histopathological images acquired from breast microscopic tissues. Extensive evaluations are carried out in terms of image classification (i.e., benign versus actionable categorization) and retrieval tests. Our framework achieves about 88.1 classification accuracy as well as promising time efficiency. For example, the framework can execute around 800 queries in only 0.01 s, comparing favorably with other commonly used dimensionality reduction and feature selection methods.",
"In this paper, we propose a new classification method for five categories of lung tissues in high-resolution computed tomography (HRCT) images, with feature-based image patch approximation. We design two new feature descriptors for higher feature descriptiveness, namely the rotation-invariant Gabor-local binary patterns (RGLBP) texture descriptor and multi-coordinate histogram of oriented gradients (MCHOG) gradient descriptor. Together with intensity features, each image patch is then labeled based on its feature approximation from reference image patches. And a new patch-adaptive sparse approximation (PASA) method is designed with the following main components: minimum discrepancy criteria for sparse-based classification, patch-specific adaptation for discriminative approximation, and feature-space weighting for distance computation. The patch-wise labelings are then accumulated as probabilistic estimations for region-level classification. The proposed method is evaluated on a publicly available ILD database, showing encouraging performance improvements over the state-of-the-arts.",
"Abstract Labeling a histopathology image as having cancerous regions or not is a critical task in cancer diagnosis; it is also clinically important to segment the cancer tissues and cluster them into various classes. Existing supervised approaches for image classification and segmentation require detailed manual annotations for the cancer pixels, which are time-consuming to obtain. In this paper, we propose a new learning method, multiple clustered instance learning (MCIL) (along the line of weakly supervised learning) for histopathology image segmentation. The proposed MCIL method simultaneously performs image-level classification (cancer vs. non-cancer image), medical image segmentation (cancer vs. non-cancer tissue), and patch-level clustering (different classes). We embed the clustering concept into the multiple instance learning (MIL) setting and derive a principled solution to performing the above three tasks in an integrated framework. In addition, we introduce contextual constraints as a prior for MCIL, which further reduces the ambiguity in MIL. Experimental results on histopathology colon cancer images and cytology images demonstrate the great advantage of MCIL over the competing methods.",
"Abstract Accurate subtyping of ovarian carcinomas is an increasingly critical and often challenging diagnostic process. This work focuses on the development of an automatic classification model for ovarian carcinoma subtyping. Specifically, we present a novel clinically inspired contextual model for histopathology image subtyping of ovarian carcinomas. A whole slide image is modelled using a collection of tissue patches extracted at multiple magnifications. An efficient and effective feature learning strategy is used for feature representation of a tissue patch. The locations of salient, discriminative tissue regions are treated as latent variables allowing the model to explicitly ignore portions of the large tissue section that are unimportant for classification. These latent variables are considered in a structured formulation to model the contextual information represented from the multi-magnification analysis of tissues. A novel, structured latent support vector machine formulation is defined and used to combine information from multiple magnifications while simultaneously operating within the latent variable framework. The structural and contextual nature of our method addresses the challenges of intra-class variation and pathologists’ workload, which are prevalent in histopathology image classification. Extensive experiments on a dataset of 133 patients demonstrate the efficacy and accuracy of the proposed method against state-of-the-art approaches for histopathology image classification. We achieve an average multi-class classification accuracy of 90 , outperforming existing works while obtaining substantial agreement with six clinicians tested on the same dataset.",
"Automatically recognizing cancers from multi-gigapixel whole slide histopathology images is one of the challenges facing machine and deep learning based solutions for digital pathology. Currently, most automatic systems for histopathology are not scalable to large images and hence require a patch-based representation; a sub-optimal solution as it results in important additional computational costs but more importantly in the loss of contextual information. We present a novel attention-based model for predicting cancer from histopathology whole slide images. The proposed model is capable of attending to the most discriminative regions of an image by adaptively selecting a limited sequence of locations and only processing the selected areas of tissues. We demonstrate the utility of the proposed model on the slide-based prediction of macro and micro metastases in sentinel lymph nodes of breast cancer patients. We achieve competitive results with state-of-the-art convolutional networks while automatically identifying discriminative areas of tissues.",
"The notion of using context information for solving high-level vision and medical image segmentation problems has been increasingly realized in the field. However, how to learn an effective and efficient context model, together with an image appearance model, remains mostly unknown. The current literature using Markov Random Fields (MRFs) and Conditional Random Fields (CRFs) often involves specific algorithm design in which the modeling and computing stages are studied in isolation. In this paper, we propose a learning algorithm, auto-context. Given a set of training images and their corresponding label maps, we first learn a classifier on local image patches. The discriminative probability (or classification confidence) maps created by the learned classifier are then used as context information, in addition to the original image patches, to train a new classifier. The algorithm then iterates until convergence. Auto-context integrates low-level and context information by fusing a large number of low-level appearance features with context and implicit shape information. The resulting discriminative algorithm is general and easy to implement. Under nearly the same parameter settings in training, we apply the algorithm to three challenging vision applications: foreground background segregation, human body configuration estimation, and scene region labeling. Moreover, context also plays a very important role in medical brain images where the anatomical structures are mostly constrained to relatively fixed positions. With only some slight changes resulting from using 3D instead of 2D features, the auto-context algorithm applied to brain MRI image segmentation is shown to outperform state-of-the-art algorithms specifically designed for this domain. Furthermore, the scope of the proposed algorithm goes beyond image analysis and it has the potential to be used for a wide variety of problems for structured prediction problems.",
"Histopathology defines the gold standard in oncology. Automatic analysis of pathology images could thus have a significant impact on diagnoses, prognoses and treatment decisions for cancer patients. Recently, convolutional neural networks (CNNs) have shown strong performance in computational histopathology tasks. However, given it remains intractable to process pathology slides in their entirety, CNNs have traditionally performed inference on small individual patches extracted from the image. This often requires a significant amount of computation and can result in ignoring potentially relevant spatial and contextual information. Being able to process larger input patches and locating discriminatory regions more efficiently could help improve both computational and task specific performance. Inspired by the recent success of Deep Recurrent Attention Models (DRAMs) in image recognition tasks, we propose a novel attention-based architecture for classification in histopathology. Similar to CNNs, DRAMs have a degree of translation invariance built-in, but the amount of computation performed can be controlled independently from the input image size. The model is a deep recurrent neural network trained with reinforcement learning to attend to the most relevant areas of large input patches. We evaluate our model on histological and molecular subtype classification tasks for the glioma cohorts of The Cancer Genome Atlas (TCGA). Our results suggest that the DRAM has comparable performance to state-of-the-art CNNs despite only processing a select number of patches.",
"Computer-aided diagnosis of histopathological images usually requires to examine all cells for accurate diagnosis. Traditional computational methods may have efficiency issues when performing cell-level analysis. In this paper, we propose a robust and scalable solution to enable such analysis in a real-time fashion. Specifically, a robust segmentation method is developed to delineate cells accurately using Gaussian-based hierarchical voting and repulsive balloon model. A large-scale image retrieval approach is also designed to examine and classify each cell of a testing image by comparing it with a massive database, e.g., half-million cells extracted from the training dataset. We evaluate this proposed framework on a challenging and important clinical use case, i.e., differentiation of two types of lung cancers (the adenocarcinoma and squamous carcinoma), using thousands of lung microscopic tissue images extracted from hundreds of patients. Our method has achieved promising accuracy and running time by searching among half-million cells .",
"Convolutional Neural Networks (CNN) are state-of-the-art models for many image classification tasks. However, to recognize cancer subtypes automatically, training a CNN on gigapixel resolution Whole Slide Tissue Images (WSI) is currently computationally impossible. The differentiation of cancer subtypes is based on cellular-level visual features observed on image patch scale. Therefore, we argue that in this situation, training a patch-level classifier on image patches will perform better than or similar to an image-level classifier. The challenge becomes how to intelligently combine patch-level classification results and model the fact that not all patches will be discriminative. We propose to train a decision fusion model to aggregate patch-level predictions given by patch-level CNNs, which to the best of our knowledge has not been shown before. Furthermore, we formulate a novel Expectation-Maximization (EM) based method that automatically locates discriminative patches robustly by utilizing the spatial relationships of patches. We apply our method to the classification of glioma and non-small-cell lung carcinoma cases into subtypes. The classification accuracy of our method is similar to the inter-observer agreement between pathologists. Although it is impossible to train CNNs on WSIs, we experimentally demonstrate using a comparable non-cancer dataset of smaller images that a patch-based CNN can outperform an image-based CNN."
]
} |
1907.09478 | 2964056701 | Digital histology images are amenable to the application of convolutional neural network (CNN) for analysis due to the sheer size of pixel data present in them. CNNs are generally used for representation learning from small image patches (e.g. 224x224) extracted from digital histology images due to computational and memory constraints. However, this approach does not incorporate high-resolution contextual information in histology images. We propose a novel way to incorporate larger context by a context-aware neural network based on images with a dimension of 1,792x1,792 pixels. The proposed framework first encodes the local representation of a histology image into high dimensional features then aggregates the features by considering their spatial organization to make a final prediction. The proposed method is evaluated for colorectal cancer grading and breast cancer classification. A comprehensive analysis of some variants of the proposed method is presented. Our method outperformed the traditional patch-based approaches, problem-specific methods, and existing context-based methods quantitatively by a margin of 3.61 . Code and dataset related information is available at this link: this https URL | Awan @cite_32 presented a method for two-tier CRC grading based on the extent of deviation of the gland from its normal shape (circular elliptical). They proposed a novel Best Alignment Metric (BAM) for this purpose. As a pre-processing step, CNN based gland segmentation was performed, followed by the calculation of BAM for each gland. For every image, average BAM was considered as a feature along with two more features inspired by BAM values. In the end, an SVM classifier was trained using this feature set for CRC grading. Our proposed method differs from these existing methods in two ways. First, it does not depend on the intermediate step of gland segmentation making it independent of segmentation inaccuracies. Second, the proposed method is entirely based on a deep neural network which makes this framework independent of cancer type. Therefore, the proposed framework could be used for other context-based histology image analysis problems. In this regard, besides CRC grading, we have demonstrated the application of the proposed method for breast cancer classification. | {
"cite_N": [
"@cite_32"
],
"mid": [
"2769999077"
],
"abstract": [
"Determining the grade of colon cancer from tissue slides is a routine part of the pathological analysis. In the case of colorectal adenocarcinoma (CRA), grading is partly determined by morphology and degree of formation of glandular structures. Achieving consistency between pathologists is difficult due to the subjective nature of grading assessment. An objective grading using computer algorithms will be more consistent, and will be able to analyse images in more detail. In this paper, we measure the shape of glands with a novel metric that we call the Best Alignment Metric (BAM). We show a strong correlation between a novel measure of glandular shape and grade of the tumour. We used shape specific parameters to perform a two-class classification of images into normal or cancerous tissue and a three-class classification into normal, low grade cancer, and high grade cancer. The task of detecting gland boundaries, which is a prerequisite of shape-based analysis, was carried out using a deep convolutional neural network designed for segmentation of glandular structures. A support vector machine (SVM) classifier was trained using shape features derived from BAM. Through cross-validation, we achieved an accuracy of 97 for the two-class and 91 for three-class classification."
]
} |
1907.09695 | 2962736944 | The problem of a deep learning model losing performance on a previously learned task when fine-tuned to a new one is a phenomenon known as Catastrophic forgetting. There are two major ways to mitigate this problem: either preserving activations of the initial network during training with a new task; or restricting the new network activations to remain close to the initial ones. The latter approach falls under the denomination of lifelong learning, where the model is updated in a way that it performs well on both old and new tasks, without having access to the old task's training samples anymore. Recently, approaches like pruning networks for freeing network capacity during sequential learning of tasks have been gaining in popularity. Such approaches allow learning small networks while making redundant parameters available for the next tasks. The common problem encountered with these approaches is that the pruning percentage is hard-coded, irrespective of the number of samples, of the complexity of the learning task and of the number of classes in the dataset. We propose a method based on Bayesian optimization to perform adaptive compression pruning of the network and show its effectiveness in lifelong learning. Our method learns to perform heavy pruning for small and or simple datasets while using milder compression rates for large and or complex data. Experiments on classification and semantic segmentation demonstrate the applicability of learning network compression, where we are able to effectively preserve performances along sequences of tasks of varying complexity. | The most common way to learn a new task from a model trained on another is to fine-tune it @cite_15 @cite_11 . Fine-tuning works generally very well for the new task, but at the price of a drop in accuracy for the former, since the weights are modified and tuned for the new task. A first possible solution is to keep a copy of the original model trained on the original task, but this leads to heavy memory requirements with an increase in the number of tasks. Another solution would be to perform multi-task learning @cite_16 , but this strategy relies on labeled data for all tasks to be available during training, which is typically not possible in sequential learning. | {
"cite_N": [
"@cite_15",
"@cite_16",
"@cite_11"
],
"mid": [
"2473930607",
"2949808626",
"2601322194"
],
"abstract": [
"When building a unified vision system or gradually adding new apabilities to a system, the usual assumption is that training data for all tasks is always available. However, as the number of tasks grows, storing and retraining on such data becomes infeasible. A new problem arises where we add new capabilities to a Convolutional Neural Network (CNN), but the training data for its existing capabilities are unavailable. We propose our Learning without Forgetting method, which uses only new task data to train the network while preserving the original capabilities. Our method performs favorably compared to commonly used feature extraction and fine-tuning adaption techniques and performs similarly to multitask learning that uses original task data we assume unavailable. A more surprising observation is that Learning without Forgetting may be able to replace fine-tuning with similar old and new task datasets for improved new task performance.",
"When building a unified vision system or gradually adding new capabilities to a system, the usual assumption is that training data for all tasks is always available. However, as the number of tasks grows, storing and retraining on such data becomes infeasible. A new problem arises where we add new capabilities to a Convolutional Neural Network (CNN), but the training data for its existing capabilities are unavailable. We propose our Learning without Forgetting method, which uses only new task data to train the network while preserving the original capabilities. Our method performs favorably compared to commonly used feature extraction and fine-tuning adaption techniques and performs similarly to multitask learning that uses original task data we assume unavailable. A more surprising observation is that Learning without Forgetting may be able to replace fine-tuning with similar old and new task datasets for improved new task performance.",
"Imitation learning has been commonly applied to solve different tasks in isolation. This usually requires either careful feature engineering, or a significant number of samples. This is far from what we desire: ideally, robots should be able to learn from very few demonstrations of any given task, and instantly generalize to new situations of the same task, without requiring task-specific engineering. In this paper, we propose a meta-learning framework for achieving such capability, which we call one-shot imitation learning. Specifically, we consider the setting where there is a very large set of tasks, and each task has many instantiations. For example, a task could be to stack all blocks on a table into a single tower, another task could be to place all blocks on a table into two-block towers, etc. In each case, different instances of the task would consist of different sets of blocks with different initial states. At training time, our algorithm is presented with pairs of demonstrations for a subset of all tasks. A neural net is trained that takes as input one demonstration and the current state (which initially is the initial state of the other demonstration of the pair), and outputs an action with the goal that the resulting sequence of states and actions matches as closely as possible with the second demonstration. At test time, a demonstration of a single instance of a new task is presented, and the neural net is expected to perform well on new instances of this new task. The use of soft attention allows the model to generalize to conditions and tasks unseen in the training data. We anticipate that by training this model on a much greater variety of tasks and settings, we will obtain a general system that can turn any demonstrations into robust policies that can accomplish an overwhelming variety of tasks. Videos available at this https URL ."
]
} |
1907.09695 | 2962736944 | The problem of a deep learning model losing performance on a previously learned task when fine-tuned to a new one is a phenomenon known as Catastrophic forgetting. There are two major ways to mitigate this problem: either preserving activations of the initial network during training with a new task; or restricting the new network activations to remain close to the initial ones. The latter approach falls under the denomination of lifelong learning, where the model is updated in a way that it performs well on both old and new tasks, without having access to the old task's training samples anymore. Recently, approaches like pruning networks for freeing network capacity during sequential learning of tasks have been gaining in popularity. Such approaches allow learning small networks while making redundant parameters available for the next tasks. The common problem encountered with these approaches is that the pruning percentage is hard-coded, irrespective of the number of samples, of the complexity of the learning task and of the number of classes in the dataset. We propose a method based on Bayesian optimization to perform adaptive compression pruning of the network and show its effectiveness in lifelong learning. Our method learns to perform heavy pruning for small and or simple datasets while using milder compression rates for large and or complex data. Experiments on classification and semantic segmentation demonstrate the applicability of learning network compression, where we are able to effectively preserve performances along sequences of tasks of varying complexity. | The issue of accessing the data of previous tasks is mitigated to a large extent in the Learning without Forgetting' (LwF) framework @cite_5 @cite_24 . LwF combines fine-tuning and distillation networks @cite_2 , where a knowledge distillation loss @cite_2 tries to preserve the output of the former classifier on data from the new task. However, LwF uses several losses, whose number (and balancing weights involved) scales linearly with the number of tasks. The authors in @cite_0 @cite_25 propose approaches where the distance between parameters of the models trained on the old and new tasks is regulated via @math losses. As for LwF, the number of parameters increases with the number of tasks. In @cite_23 , the authors use autoencoders in addition to LwF. This approach has an overhead of a linearly increasing number of autoencoders and task-specific classifiers, several hyperparameters and also a distillation loss between the single-task and the multitask model, making its training complex. | {
"cite_N": [
"@cite_24",
"@cite_0",
"@cite_23",
"@cite_2",
"@cite_5",
"@cite_25"
],
"mid": [
"2786498526",
"2737691244",
"2131479143",
"2605911906",
"2964244278",
"2554616628"
],
"abstract": [
"In this paper, we address the incremental classifier learning problem, which suffers from catastrophic forgetting. The main reason for catastrophic forgetting is that the past data are not available during learning. Typical approaches keep some exemplars for the past classes and use distillation regularization to retain the classification capability on the past classes and balance the past and new classes. However, there are four main problems with these approaches. First, the loss function is not efficient for classification. Second, there is unbalance problem between the past and new classes. Third, the size of pre-decided exemplars is usually limited and they might not be distinguishable from unseen new classes. Forth, the exemplars may not be allowed to be kept for a long time due to privacy regulations. To address these problems, we propose (a) a new loss function to combine the cross-entropy loss and distillation loss, (b) a simple way to estimate and remove the unbalance between the old and new classes , and (c) using Generative Adversarial Networks (GANs) to generate historical data and select representative exemplars during generation. We believe that the data generated by GANs have much less privacy issues than real images because GANs do not directly copy any real image patches. We evaluate the proposed method on CIFAR-100, Flower-102, and MS-Celeb-1M-Base datasets and extensive experiments demonstrate the effectiveness of our method.",
"In this paper, we study the problem of training large-scale face identification model with imbalanced training data. This problem naturally exists in many real scenarios including large-scale celebrity recognition, movie actor annotation, etc. Our solution contains two components. First, we build a face feature extraction model, and improve its performance, especially for the persons with very limited training samples, by introducing a regularizer to the cross entropy loss for the multi-nomial logistic regression (MLR) learning. This regularizer encourages the directions of the face features from the same class to be close to the direction of their corresponding classification weight vector in the logistic regression. Second, we build a multi-class classifier using MLR on top of the learned face feature extraction model. Since the standard MLR has poor generalization capability for the one-shot classes even if these classes have been oversampled, we propose a novel supervision signal called underrepresented-classes promotion loss, which aligns the norms of the weight vectors of the one-shot classes (a.k.a. underrepresented-classes) to those of the normal classes. In addition to the original cross entropy loss, this new loss term effectively promotes the underrepresented classes in the learned model and leads to a remarkable improvement in face recognition performance. We test our solution on the MS-Celeb-1M low-shot learning benchmark task. Our solution recognizes 94.89 of the test images at the precision of 99 for the one-shot classes. To the best of our knowledge, this is the best performance among all the published methods using this benchmark task with the same setup, including all the participants in the recent MS-Celeb-1M challenge at ICCV 2017.",
"Consider the problem of learning logistic-regression models for multiple classification tasks, where the training data set for each task is not drawn from the same statistical distribution. In such a multi-task learning (MTL) scenario, it is necessary to identify groups of similar tasks that should be learned jointly. Relying on a Dirichlet process (DP) based statistical model to learn the extent of similarity between classification tasks, we develop computationally efficient algorithms for two different forms of the MTL problem. First, we consider a symmetric multi-task learning (SMTL) situation in which classifiers for multiple tasks are learned jointly using a variational Bayesian (VB) algorithm. Second, we consider an asymmetric multi-task learning (AMTL) formulation in which the posterior density function from the SMTL model parameters (from previous tasks) is used as a prior for a new task: this approach has the significant advantage of not requiring storage and use of all previous data from prior tasks. The AMTL formulation is solved with a simple Markov Chain Monte Carlo (MCMC) construction. Experimental results on two real life MTL problems indicate that the proposed algorithms: (a) automatically identify subgroups of related tasks whose training data appear to be drawn from similar distributions; and (b) are more accurate than simpler approaches such as single-task learning, pooling of data across all tasks, and simplified approximations to DP.",
"This paper introduces a new lifelong learning solution where a single model is trained for a sequence of tasks. The main challenge that vision systems face in this context is catastrophic forgetting: as they tend to adapt to the most recently seen task, they lose performance on the tasks that were learned previously. Our method aims at preserving the knowledge of the previous tasks while learning a new one by using autoencoders. For each task, an under-complete autoencoder is learned, capturing the features that are crucial for its achievement. When a new task is presented to the system, we prevent the reconstructions of the features with these autoencoders from changing, which has the effect of preserving the information on which the previous tasks are mainly relying. At the same time, the features are given space to adjust to the most recent environment as only their projection into a low dimension submanifold is controlled. The proposed system is evaluated on image classification tasks and shows a reduction of forgetting over the state-of-the-art",
"This paper introduces a new lifelong learning solution where a single model is trained for a sequence of tasks. The main challenge that vision systems face in this context is catastrophic forgetting: as they tend to adapt to the most recently seen task, they lose performance on the tasks that were learned previously. Our method aims at preserving the knowledge of the previous tasks while learning a new one by using autoencoders. For each task, an under-complete autoencoder is learned, capturing the features that are crucial for its achievement. When a new task is presented to the system, we prevent the reconstructions of the features with these autoencoders from changing, which has the effect of preserving the information on which the previous tasks are mainly relying. At the same time, the features are given space to adjust to the most recent environment as only their projection into a low dimension submanifold is controlled. The proposed system is evaluated on image classification tasks and shows a reduction of forgetting over the state-ofthe-art.",
"In this paper we introduce a model of lifelong learning, based on a Network of Experts. New tasks experts are learned and added to the model sequentially, building on what was learned before. To ensure scalability of this process, data from previous tasks cannot be stored and hence is not available when learning a new task. A critical issue in such context, not addressed in the literature so far, relates to the decision which expert to deploy at test time. We introduce a set of gating autoencoders that learn a representation for the task at hand, and, at test time, automatically forward the test sample to the relevant expert. This also brings memory efficiency as only one expert network has to be loaded into memory at any given time. Further, the autoencoders inherently capture the relatedness of one task to another, based on which the most relevant prior model to be used for training a new expert, with fine-tuning or learning-without-forgetting, can be selected. We evaluate our method on image classification and video prediction problems."
]
} |
1907.09695 | 2962736944 | The problem of a deep learning model losing performance on a previously learned task when fine-tuned to a new one is a phenomenon known as Catastrophic forgetting. There are two major ways to mitigate this problem: either preserving activations of the initial network during training with a new task; or restricting the new network activations to remain close to the initial ones. The latter approach falls under the denomination of lifelong learning, where the model is updated in a way that it performs well on both old and new tasks, without having access to the old task's training samples anymore. Recently, approaches like pruning networks for freeing network capacity during sequential learning of tasks have been gaining in popularity. Such approaches allow learning small networks while making redundant parameters available for the next tasks. The common problem encountered with these approaches is that the pruning percentage is hard-coded, irrespective of the number of samples, of the complexity of the learning task and of the number of classes in the dataset. We propose a method based on Bayesian optimization to perform adaptive compression pruning of the network and show its effectiveness in lifelong learning. Our method learns to perform heavy pruning for small and or simple datasets while using milder compression rates for large and or complex data. Experiments on classification and semantic segmentation demonstrate the applicability of learning network compression, where we are able to effectively preserve performances along sequences of tasks of varying complexity. | An alternative direction to the above is the idea of removing redundant parameters by neural network compression @cite_18 . The authors report good results but only use a fixed pruning percentage for all the tasks, irrespective of the complexity of the data involved. Other works have used masks on networks' weights, either using attention @cite_27 or by learning binary weights masks end-to-end, in order to use only the weights useful for the new task @cite_6 . | {
"cite_N": [
"@cite_27",
"@cite_18",
"@cite_6"
],
"mid": [
"2788715907",
"2791091755",
"2964019666"
],
"abstract": [
"In recent years considerable research efforts have been devoted to compression techniques of convolutional neural networks (CNNs). Many works so far have focused on CNN connection pruning methods which produce sparse parameter tensors in convolutional or fully-connected layers. It has been demonstrated in several studies that even simple methods can effectively eliminate connections of a CNN. However, since these methods make parameter tensors just sparser but no smaller, the compression may not transfer directly to acceleration without support from specially designed hardware. In this paper, we propose an iterative approach named Auto-balanced Filter Pruning, where we pre-train the network in an innovative auto-balanced way to transfer the representational capacity of its convolutional layers to a fraction of the filters, prune the redundant ones, then re-train it to restore the accuracy. In this way, a smaller version of the original network is learned and the floating-point operations (FLOPs) are reduced. By applying this method on several common CNNs, we show that a large portion of the filters can be discarded without obvious accuracy drop, leading to significant reduction of computational burdens. Concretely, we reduce the inference cost of LeNet-5 on MNIST, VGG-16 and ResNet-56 on CIFAR-10 by 95.1 , 79.7 and 60.9 , respectively. Copyright © 2018, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.",
"This work presents a method for adapting a single, fixed deep neural network to multiple tasks without affecting performance on already learned tasks. By building upon ideas from network quantization and pruning, we learn binary masks that “piggyback” on an existing network, or are applied to unmodified weights of that network to provide good performance on a new task. These masks are learned in an end-to-end differentiable fashion, and incur a low overhead of 1 bit per network parameter, per task. Even though the underlying network is fixed, the ability to mask individual weights allows for the learning of a large number of filters. We show performance comparable to dedicated fine-tuned networks for a variety of classification tasks, including those with large domain shifts from the initial task (ImageNet), and a variety of network architectures. Our performance is agnostic to task ordering and we do not suffer from catastrophic forgetting or competition between tasks.",
"Recently there has been a lot of work on pruning filters from deep convolutional neural networks (CNNs) with the intention of reducing computations. The key idea is to rank the filters based on a certain criterion (say, l1-norm, average percentage of zeros, etc) and retain only the top ranked filters. Once the low scoring filters are pruned away the remainder of the network is fine tuned and is shown to give performance comparable to the original unpruned network. In this work, we report experiments which suggest that the comparable performance of the pruned network is not due to the specific criterion chosen but due to the inherent plasticity of deep neural networks which allows them to recover from the loss of pruned filters once the rest of the filters are fine-tuned. Specifically, we show counter-intuitive results wherein by randomly pruning 25-50 filters from deep CNNs we are able to obtain the same performance as obtained by using state of the art pruning methods. We empirically validate our claims by doing an exhaustive evaluation with VGG-16 and ResNet-50. Further, we also evaluate a real world scenario where a CNN trained on all 1000 ImageNet classes needs to be tested on only a small set of classes at test time (say, only animals). We create a new benchmark dataset from ImageNet to evaluate such class specific pruning and show that even here a random pruning strategy gives close to state of the art performance. Lastly, unlike existing approaches which mainly focus on the task of image classification, in this work we also report results on object detection. We show that using a simple random pruning strategy we can achieve significant speed up in object detection (74 improvement in fps) while retaining the same accuracy as that of the original Faster RCNN model."
]
} |
1907.09595 | 2963037463 | Depthwise convolution is becoming increasingly popular in modern efficient ConvNets, but its kernel size is often overlooked. In this paper, we systematically study the impact of different kernel sizes, and observe that combining the benefits of multiple kernel sizes can lead to better accuracy and efficiency. Based on this observation, we propose a new mixed depthwise convolution (MixConv), which naturally mixes up multiple kernel sizes in a single convolution. As a simple drop-in replacement of vanilla depthwise convolution, our MixConv improves the accuracy and efficiency for existing MobileNets on both ImageNet classification and COCO object detection. To demonstrate the effectiveness of MixConv, we integrate it into AutoML search space and develop a new family of models, named as MixNets, which outperform previous mobile models including MobileNetV2 [20] (ImageNet top-1 accuracy +4.2 ), ShuffleNetV2 [16] (+3.5 ), MnasNet [26] (+1.3 ), ProxylessNAS [2] (+2.2 ), and FBNet [27] (+2.0 ). In particular, our MixNet-L achieves a new state-of-the-art 78.9 ImageNet top-1 accuracy under typical mobile settings (<600M FLOPS). Code is at this https URL tensorflow tpu tree master models official mnasnet mixnet | In recent years, significant efforts have been spent on improving ConvNet efficiency, from more efficient convolutional operations @cite_30 @cite_5 @cite_32 , bottleneck layers @cite_15 @cite_0 , to more efficient architectures @cite_17 @cite_26 @cite_13 . In particular, depthwise convolution has been increasingly popular in all mobile-size ConvNets, such as MobileNets @cite_5 @cite_15 , ShuffleNets @cite_20 @cite_28 , MnasNet @cite_17 , and beyond @cite_21 @cite_6 @cite_4 . Recently, EfficientNet @cite_27 even achieves both state-of-the-art ImageNet accuracy and ten-fold better efficiency by extensively using depthwise and pointwise convolutions. Unlike regular convolution, depthwise convolution performs convolutional kernels for each channel separately, thus reducing parameter size and computational cost. Our proposed MixConv generalizes the concept of depthwise convolution, and can be considered as a drop-in replacement of vanilla depthwise convolution. | {
"cite_N": [
"@cite_30",
"@cite_26",
"@cite_4",
"@cite_28",
"@cite_21",
"@cite_32",
"@cite_6",
"@cite_0",
"@cite_27",
"@cite_5",
"@cite_15",
"@cite_13",
"@cite_20",
"@cite_17"
],
"mid": [
"2962835968",
"1686810756",
"2885340141",
"2963342610",
"2581531918",
"2938458886",
"2177847924",
"2125085157",
"2962988160",
"2952713184",
"2963446712",
"2511730936",
"2591237478",
"2513928851"
],
"abstract": [
"Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.",
"In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.",
"Depthwise separable convolution has shown great efficiency in network design, but requires time-consuming training procedure with full training-set available. This paper first analyzes the mathematical relationship between regular convolutions and depthwise separable convolutions, and proves that the former one could be approximated with the latter one in closed form. We show depthwise separable convolutions are principal components of regular convolutions. And then we propose network decoupling (ND), a training-free method to accelerate convolutional neural networks (CNNs) by transferring pre-trained CNN models into the MobileNet-like depthwise separable convolution structure, with a promising speedup yet negligible accuracy loss. We further verify through experiments that the proposed method is orthogonal to other training-free methods like channel decomposition, spatial decomposition, etc. Combining the proposed method with them will bring even larger CNN speedup. For instance, ND itself achieves about 2X speedup for the widely used VGG16, and combined with other methods, it reaches 3.7X speedup with graceful accuracy degradation. We demonstrate that ND is widely applicable to classification networks like ResNet, and object detection network like SSD300.",
"An increasing need of running Convolutional Neural Network (CNN) models on mobile devices with limited computing power and memory resource encourages studies on efficient model design. A number of efficient architectures have been proposed in recent years, for example, MobileNet, ShuffleNet, and NASNet-A. However, all these models are heavily dependent on depthwise separable convolution which lacks efficient implementation in most deep learning frameworks. In this study, we propose an efficient architecture named PeleeNet, which is built with conventional convolution instead. On ImageNet ILSVRC 2012 dataset, our proposed PeleeNet achieves a higher accuracy by 1.8 (72.4 vs. 70.6 ) and 23 faster speed than MobileNet, the state-of-the-art efficient architecture. Meanwhile, PeleeNet is only 66 of the model size of MobileNet. We then propose a real-time object detection system by combining PeleeNet with Single Shot MultiBox Detector (SSD) method and optimizing the architecture for fast speed. Our proposed detection system, named Pelee, achieves 76.4 mAP (mean average precision) on PASCAL VOC2007 and 22.4 mAP on MS COCO dataset at the speed of 23.6 FPS on iPhone 8 and 74 FPS on NVIDIA TX2. The result on COCO outperforms YOLOv2 in consideration of a higher precision, 13.6 times lower computational cost and 11.3 times smaller model size. The code and models are open sourced.",
"Deep convolutional neural networks achieve remarkable visual recognition performance, at the cost of high computational complexity. In this paper, we have a new design of efficient convolutional layers based on three schemes. The 3D convolution operation in a convolutional layer can be considered as performing spatial convolution in each channel and linear projection across channels simultaneously. By unravelling them and arranging the spatial convolution sequentially, the proposed layer is composed of a single intra-channel convolution, of which the computation is negligible, and a linear channel projection. A topological subdivisioning is adopted to reduce the connection between the input channels and output channels. Additionally, we also introduce a spatial \"bottleneck\" structure that utilizes a convolution-projection-deconvolution pipeline to take advantage of the correlation between adjacent pixels in the input. Our experiments demonstrate that the proposed layers remarkably outperform the standard convolutional layers with regard to accuracy complexity ratio. Our models achieve similar accuracy to VGG, ResNet-50, ResNet-101 while requiring 42, 4.5, 6.5 times less computation respectively.",
"In natural images, information is conveyed at different frequencies where higher frequencies are usually encoded with fine details and lower frequencies are usually encoded with global structures. Similarly, the output feature maps of a convolution layer can also be seen as a mixture of information at different frequencies. In this work, we propose to factorize the mixed feature maps by their frequencies and design a novel Octave Convolution (OctConv) operation to store and process feature maps that vary spatially \"slower\" at a lower spatial resolution reducing both memory and computation cost. Unlike existing multi-scale meth-ods, OctConv is formulated as a single, generic, plug-and-play convolutional unit that can be used as a direct replacement of (vanilla) convolutions without any adjustments in the network architecture. It is also orthogonal and complementary to methods that suggest better topologies or reduce channel-wise redundancy like group or depth-wise convolutions. We experimentally show that by simply replacing con-volutions with OctConv, we can consistently boost accuracy for both image and video recognition tasks, while reducing memory and computational cost. An OctConv-equipped ResNet-152 can achieve 82.9 top-1 classification accuracy on ImageNet with merely 22.2 GFLOPs.",
"Although the latest high-end smartphone has powerful CPU and GPU, running deeper convolutional neural networks (CNNs) for complex tasks such as ImageNet classification on mobile devices is challenging. To deploy deep CNNs on mobile devices, we present a simple and effective scheme to compress the entire CNN, which we call one-shot whole network compression. The proposed scheme consists of three steps: (1) rank selection with variational Bayesian matrix factorization, (2) Tucker decomposition on kernel tensor, and (3) fine-tuning to recover accumulated loss of accuracy, and each step can be easily implemented using publicly available tools. We demonstrate the effectiveness of the proposed scheme by testing the performance of various compressed CNNs (AlexNet, VGGS, GoogLeNet, and VGG-16) on the smartphone. Significant reductions in model size, runtime, and energy consumption are obtained, at the cost of small loss in accuracy. In addition, we address the important implementation level issue on 1?1 convolution, which is a key operation of inception module of GoogLeNet as well as CNNs compressed by our proposed scheme.",
"We apply Convolutional Networks (ConvNets) to the task of traffic sign classification as part of the GTSRB competition. ConvNets are biologically-inspired multi-stage architectures that automatically learn hierarchies of invariant features. While many popular vision approaches use hand-crafted features such as HOG or SIFT, ConvNets learn features at every level from data that are tuned to the task at hand. The traditional ConvNet architecture was modified by feeding 1st stage features in addition to 2nd stage features to the classifier. The system yielded the 2nd-best accuracy of 98.97 during phase I of the competition (the best entry obtained 98.98 ), above the human performance of 98.81 , using 32×32 color input images. Experiments conducted after phase 1 produced a new record of 99.17 by increasing the network capacity, and by using greyscale images instead of color. Interestingly, random features still yielded competitive results (97.33 ).",
"Abstract: Although the latest high-end smartphone has powerful CPU and GPU, running deeper convolutional neural networks (CNNs) for complex tasks such as ImageNet classification on mobile devices is challenging. To deploy deep CNNs on mobile devices, we present a simple and effective scheme to compress the entire CNN, which we call one-shot whole network compression. The proposed scheme consists of three steps: (1) rank selection with variational Bayesian matrix factorization, (2) Tucker decomposition on kernel tensor, and (3) fine-tuning to recover accumulated loss of accuracy, and each step can be easily implemented using publicly available tools. We demonstrate the effectiveness of the proposed scheme by testing the performance of various compressed CNNs (AlexNet, VGGS, GoogLeNet, and VGG-16) on the smartphone. Significant reductions in model size, runtime, and energy consumption are obtained, at the cost of small loss in accuracy. In addition, we address the important implementation level issue on 1?1 convolution, which is a key operation of inception module of GoogLeNet as well as CNNs compressed by our proposed scheme.",
"Deep convolutional neural networks (ConvNets) of 3-dimensional kernels allow joint modeling of spatiotemporal features. These networks have improved performance of video and volumetric image analysis, but have been limited in size due to the low memory ceiling of GPU hardware. Existing CPU implementations overcome this constraint but are impractically slow. Here we extend and optimize the faster Winograd-class of convolutional algorithms to the @math -dimensional case and specifically for CPU hardware. First, we remove the need to manually hand-craft algorithms by exploiting the relaxed constraints and cheap sparse access of CPU memory. Second, we maximize CPU utilization and multicore scalability by transforming data matrices to be cache-aware, integer multiples of AVX vector widths. Treating 2-dimensional ConvNets as a special (and the least beneficial) case of our approach, we demonstrate a 5 to 25-fold improvement in throughput compared to previous state-of-the-art.",
"Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output. In this paper, we embrace this observation and introduce the Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion. Whereas traditional convolutional networks with L layers have L connections—one between each layer and its subsequent layer—our network has L(L+1) 2 direct connections. For each layer, the feature-maps of all preceding layers are used as inputs, and its own feature-maps are used as inputs into all subsequent layers. DenseNets have several compelling advantages: they alleviate the vanishing-gradient problem, strengthen feature propagation, encourage feature reuse, and substantially reduce the number of parameters. We evaluate our proposed architecture on four highly competitive object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet). DenseNets obtain significant improvements over the state-of-the-art on most of them, whilst requiring less memory and computation to achieve high performance. Code and pre-trained models are available at https: github.com liuzhuang13 DenseNet.",
"Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output. In this paper, we embrace this observation and introduce the Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion. Whereas traditional convolutional networks with L layers have L connections - one between each layer and its subsequent layer - our network has L(L+1) 2 direct connections. For each layer, the feature-maps of all preceding layers are used as inputs, and its own feature-maps are used as inputs into all subsequent layers. DenseNets have several compelling advantages: they alleviate the vanishing-gradient problem, strengthen feature propagation, encourage feature reuse, and substantially reduce the number of parameters. We evaluate our proposed architecture on four highly competitive object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet). DenseNets obtain significant improvements over the state-of-the-art on most of them, whilst requiring less computation to achieve high performance. Code and pre-trained models are available at this https URL .",
"Abstract Convolutional networks (ConvNets) have become a popular approach to computer vision. Here we consider the parallelization of ConvNet training, which is computationally costly. Our novel parallel algorithm is based on decomposition into a set of tasks, most of which are convolutions or FFTs. Theoretical analysis suggests that linear speedup with the number of processors is attainable. To attain such performance on real shared-memory machines, our algorithm computes convolutions converging on the same node of the network with temporal locality to reduce cache misses, and sums the convergent convolution outputs via an almost wait-free concurrent method to reduce time spent in critical sections. Benchmarking with multi-core CPUs shows speedup roughly equal to the number of physical cores. We also demonstrate 90x speedup on a many-core CPU (Xeon Phi Knights Corner). Our algorithm can be either faster or slower than certain GPU implementations depending on specifics of the network architecture, kernel sizes, and density and size of the output patch.",
"Deep convolutional neural networks achieve better than human level visual recognition accuracy, at the cost of high computational complexity. We propose to factorize the convolutional layers to improve their efficiency. In traditional convolutional layers, the 3D convolution can be considered as performing in-channel spatial convolution and linear channel projection simultaneously, leading to highly redundant computation. By unravelling them apart, the proposed layer only involves single in-channel convolution and linear channel projection. When stacking such layers together, we achieves similar accuracy with significantly less computation. Additionally, we propose a topological connection framework between the input channels and output channels that further improves the layer's efficiency. Our experiments demonstrate that the proposed method remarkably outperforms the standard convolutional layer with regard to accuracy complexity ratio. Our model achieves accuracy of GoogLeNet while consuming 3.4 times less computation."
]
} |
1907.09595 | 2963037463 | Depthwise convolution is becoming increasingly popular in modern efficient ConvNets, but its kernel size is often overlooked. In this paper, we systematically study the impact of different kernel sizes, and observe that combining the benefits of multiple kernel sizes can lead to better accuracy and efficiency. Based on this observation, we propose a new mixed depthwise convolution (MixConv), which naturally mixes up multiple kernel sizes in a single convolution. As a simple drop-in replacement of vanilla depthwise convolution, our MixConv improves the accuracy and efficiency for existing MobileNets on both ImageNet classification and COCO object detection. To demonstrate the effectiveness of MixConv, we integrate it into AutoML search space and develop a new family of models, named as MixNets, which outperform previous mobile models including MobileNetV2 [20] (ImageNet top-1 accuracy +4.2 ), ShuffleNetV2 [16] (+3.5 ), MnasNet [26] (+1.3 ), ProxylessNAS [2] (+2.2 ), and FBNet [27] (+2.0 ). In particular, our MixNet-L achieves a new state-of-the-art 78.9 ImageNet top-1 accuracy under typical mobile settings (<600M FLOPS). Code is at this https URL tensorflow tpu tree master models official mnasnet mixnet | Our idea shares a lot of similarities to prior multi-branch ConvNets, such as Inceptions @cite_1 @cite_7 , Inception-ResNet @cite_31 , ResNeXt @cite_0 , and NASNet @cite_6 . By using multiple branches in each layer, these ConvNets are able to utilize different operations (such as convolution and pooling) in a single layer. Similarly, there are also many prior work on combining multi-scale feature maps from different layers, such as DenseNet @cite_12 @cite_3 and feature pyramid network @cite_10 . However, unlike these prior works that mostly focus on changing the macro-architecture of neural networks in order to utilize different convolutional ops, our work aims to design a drop-in replacement of a single depthwise convolution, with the goal of easily utilizing different kernel sizes without changing the network structure. | {
"cite_N": [
"@cite_7",
"@cite_1",
"@cite_6",
"@cite_3",
"@cite_0",
"@cite_31",
"@cite_10",
"@cite_12"
],
"mid": [
"2757338536",
"2884751099",
"2531409750",
"2951583185",
"2258484932",
"2962742544",
"2785325870",
"2109255472"
],
"abstract": [
"While much of the work in the design of convolutional networks over the last five years has revolved around the empirical investigation of the importance of depth, filter sizes, and number of feature channels, recent studies have shown that branching, i.e., splitting the computation along parallel but distinct threads and then aggregating their outputs, represents a new promising dimension for significant improvements in performance. To combat the complexity of design choices in multi-branch architectures, prior work has adopted simple strategies, such as a fixed branching factor, the same input being fed to all parallel branches, and an additive combination of the outputs produced by all branches at aggregation points. In this work we remove these predefined choices and propose an algorithm to learn the connections between branches in the network. Instead of being chosen a priori by the human designer, the multi-branch connectivity is learned simultaneously with the weights of the network by optimizing a single loss function defined with respect to the end task. We demonstrate our approach on the problem of multi-class image classification using four different datasets where it yields consistently higher accuracy compared to the state-of-the-art ResNeXt'' multi-branch network given the same learning capacity.",
"Do convolutional networks really need a fixed feed-forward structure? What if, after identifying the high-level concept of an image, a network could move directly to a layer that can distinguish fine-grained differences? Currently, a network would first need to execute sometimes hundreds of intermediate layers that specialize in unrelated aspects. Ideally, the more a network already knows about an image, the better it should be at deciding which layer to compute next. In this work, we propose convolutional networks with adaptive inference graphs (ConvNet-AIG) that adaptively define their network topology conditioned on the input image. Following a high-level structure similar to residual networks (ResNets), ConvNet-AIG decides for each input image on the fly which layers are needed. In experiments on ImageNet we show that ConvNet-AIG learns distinct inference graphs for different categories. Both ConvNet-AIG with 50 and 101 layers outperform their ResNet counterpart, while using (20 ) and (33 ) less computations respectively. By grouping parameters into layers for related classes and only executing relevant layers, ConvNet-AIG improves both efficiency and overall classification quality. Lastly, we also study the effect of adaptive inference graphs on the susceptibility towards adversarial examples. We observe that ConvNet-AIG shows a higher robustness than ResNets, complementing other known defense mechanisms.",
"We present an interpretation of Inception modules in convolutional neural networks as being an intermediate step in-between regular convolution and the depthwise separable convolution operation (a depthwise convolution followed by a pointwise convolution). In this light, a depthwise separable convolution can be understood as an Inception module with a maximally large number of towers. This observation leads us to propose a novel deep convolutional neural network architecture inspired by Inception, where Inception modules have been replaced with depthwise separable convolutions. We show that this architecture, dubbed Xception, slightly outperforms Inception V3 on the ImageNet dataset (which Inception V3 was designed for), and significantly outperforms Inception V3 on a larger image classification dataset comprising 350 million images and 17,000 classes. Since the Xception architecture has the same number of parameters as Inception V3, the performance gains are not due to increased capacity but rather to a more efficient use of model parameters.",
"We present an interpretation of Inception modules in convolutional neural networks as being an intermediate step in-between regular convolution and the depthwise separable convolution operation (a depthwise convolution followed by a pointwise convolution). In this light, a depthwise separable convolution can be understood as an Inception module with a maximally large number of towers. This observation leads us to propose a novel deep convolutional neural network architecture inspired by Inception, where Inception modules have been replaced with depthwise separable convolutions. We show that this architecture, dubbed Xception, slightly outperforms Inception V3 on the ImageNet dataset (which Inception V3 was designed for), and significantly outperforms Inception V3 on a larger image classification dataset comprising 350 million images and 17,000 classes. Since the Xception architecture has the same number of parameters as Inception V3, the performance gains are not due to increased capacity but rather to a more efficient use of model parameters.",
"Convolutional neural network (CNN) has achieved the state-of-the-art performance in many different visual tasks. Learned from a large-scale training data set, CNN features are much more discriminative and accurate than the handcrafted features. Moreover, CNN features are also transferable among different domains. On the other hand, traditional dictionary-based features (such as BoW and spatial pyramid matching) contain much more local discriminative and structural information, which is implicitly embedded in the images. To further improve the performance, in this paper, we propose to combine CNN with dictionary-based models for scene recognition and visual domain adaptation (DA). Specifically, based on the well-tuned CNN models (e.g., AlexNet and VGG Net), two dictionary-based representations are further constructed, namely, mid-level local representation (MLR) and convolutional Fisher vector (CFV) representation. In MLR, an efficient two-stage clustering method, i.e., weighted spatial and feature space spectral clustering on the parts of a single image followed by clustering all representative parts of all images, is used to generate a class-mixture or a class-specific part dictionary. After that, the part dictionary is used to operate with the multiscale image inputs for generating mid-level representation. In CFV, a multiscale and scale-proportional Gaussian mixture model training strategy is utilized to generate Fisher vectors based on the last convolutional layer of CNN. By integrating the complementary information of MLR, CFV, and the CNN features of the fully connected layer, the state-of-the-art performance can be achieved on scene recognition and DA problems. An interested finding is that our proposed hybrid representation (from VGG net trained on ImageNet) is also complementary to GoogLeNet and or VGG-11 (trained on Place205) greatly.",
"Over the last years, deep convolutional neural networks (ConvNets) have transformed the field of computer vision thanks to their unparalleled capacity to learn high level semantic image features. However, in order to successfully learn those features, they usually require massive amounts of manually labeled data, which is both expensive and impractical to scale. Therefore, unsupervised semantic feature learning, i.e., learning without requiring manual annotation effort, is of crucial importance in order to successfully harvest the vast amount of visual data that are available today. In our work we propose to learn image features by training ConvNets to recognize the 2d rotation that is applied to the image that it gets as input. We demonstrate both qualitatively and quantitatively that this apparently simple task actually provides a very powerful supervisory signal for semantic feature learning. We exhaustively evaluate our method in various unsupervised feature learning benchmarks and we exhibit in all of them state-of-the-art performance. Specifically, our results on those benchmarks demonstrate dramatic improvements w.r.t. prior state-of-the-art approaches in unsupervised representation learning and thus significantly close the gap with supervised feature learning. For instance, in PASCAL VOC 2007 detection task our unsupervised pre-trained AlexNet model achieves the state-of-the-art (among unsupervised methods) mAP of 54.4 that is only 2.4 points lower from the supervised case. We get similar striking results when we transfer our unsupervised learned features on various other tasks, such as ImageNet classification, PASCAL classification, PASCAL segmentation, and CIFAR-10 classification.",
"Over the last years, deep convolutional neural networks (ConvNets) have transformed the field of computer vision thanks to their unparalleled capacity to learn high level semantic image features. However, in order to successfully learn those features, they usually require massive amounts of manually labeled data, which is both expensive and impractical to scale. Therefore, unsupervised semantic feature learning, i.e., learning without requiring manual annotation effort, is of crucial importance in order to successfully harvest the vast amount of visual data that are available today. In our work we propose to learn image features by training ConvNets to recognize the 2d rotation that is applied to the image that it gets as input. We demonstrate both qualitatively and quantitatively that this apparently simple task actually provides a very powerful supervisory signal for semantic feature learning. We exhaustively evaluate our method in various unsupervised feature learning benchmarks and we exhibit in all of them state-of-the-art performance. Specifically, our results on those benchmarks demonstrate dramatic improvements w.r.t. prior state-of-the-art approaches in unsupervised representation learning and thus significantly close the gap with supervised feature learning. For instance, in PASCAL VOC 2007 detection task our unsupervised pre-trained AlexNet model achieves the state-of-the-art (among unsupervised methods) mAP of 54.4 that is only 2.4 points lower from the supervised case. We get similarly striking results when we transfer our unsupervised learned features on various other tasks, such as ImageNet classification, PASCAL classification, PASCAL segmentation, and CIFAR-10 classification. The code and models of our paper will be published on: this https URL .",
"Existing deep convolutional neural networks (CNNs) require a fixed-size (e.g., 224 @math 224) input image. This requirement is “artificial” and may reduce the recognition accuracy for the images or sub-images of an arbitrary size scale. In this work, we equip the networks with another pooling strategy, “spatial pyramid pooling”, to eliminate the above requirement. The new network structure, called SPP-net, can generate a fixed-length representation regardless of image size scale. Pyramid pooling is also robust to object deformations. With these advantages, SPP-net should in general improve all CNN-based image classification methods. On the ImageNet 2012 dataset, we demonstrate that SPP-net boosts the accuracy of a variety of CNN architectures despite their different designs. On the Pascal VOC 2007 and Caltech101 datasets, SPP-net achieves state-of-the-art classification results using a single full-image representation and no fine-tuning. The power of SPP-net is also significant in object detection. Using SPP-net, we compute the feature maps from the entire image only once, and then pool features in arbitrary regions (sub-images) to generate fixed-length representations for training the detectors. This method avoids repeatedly computing the convolutional features. In processing test images, our method is 24-102 @math faster than the R-CNN method, while achieving better or comparable accuracy on Pascal VOC 2007. In ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2014, our methods rank #2 in object detection and #3 in image classification among all 38 teams. This manuscript also introduces the improvement made for this competition."
]
} |
1907.09705 | 2962957458 | Scene text recognition has been an important, active research topic in computer vision for years. Previous approaches mainly consider text as 1D signals and cast scene text recognition as a sequence prediction problem, by feat of CTC or attention based encoder-decoder framework, which is originally designed for speech recognition. However, different from speech voices, which are 1D signals, text instances are essentially distributed in 2D image spaces. To adhere to and make use of the 2D nature of text for higher recognition accuracy, we extend the vanilla CTC model to a second dimension, thus creating 2D-CTC. 2D-CTC can adaptively concentrate on most relevant features while excluding the impact from clutters and noises in the background; It can also naturally handle text instances with various forms (horizontal, oriented and curved) while giving more interpretable intermediate predictions. The experiments on standard benchmarks for scene text recognition, such as IIIT-5K, ICDAR 2015, SVP-Perspective, and CUTE80, demonstrate that the proposed 2D-CTC model outperforms state-of-the-art methods on the text of both regular and irregular shapes. Moreover, 2D-CTC exhibits its superiority over prior art on training and testing speed. Our implementation and models of 2D-CTC will be made publicly available soon later. | Another admirable direction of frame-wise prediction alignment is attention-based sequence encoder and decoder framework @cite_24 @cite_5 @cite_22 @cite_36 @cite_0 @cite_32 @cite_37 . The models focus on one position and predict the corresponding character at each time step, but suffer from the problems of misalignment and attention drift @cite_24 . Miss-classification at previous steps may lead to drifted attention locations and wrong predictions in successive time steps because of the recursive mechanism. Recent works take attention decoders forward to even better accuracy by suggesting new loss functions @cite_24 @cite_38 and introducing image rectification modules @cite_36 @cite_40 . Both of them bring appreciable improvement to attention decoders. However, despite the high accuracy attention decoder has achieved, the considerably lower inference speed is the fundamental factor which has limited its application in real-world text recognition systems. Detailed experiments are presented in Sec . | {
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_22",
"@cite_36",
"@cite_32",
"@cite_24",
"@cite_0",
"@cite_40",
"@cite_5"
],
"mid": [
"2952470929",
"2798484463",
"2963327605",
"2896588340",
"2724346673",
"2952469094",
"2475662749",
"2964117975",
"2500355674"
],
"abstract": [
"Recently, there has been an increasing interest in end-to-end speech recognition that directly transcribes speech to text without any predefined alignments. One approach is the attention-based encoder-decoder framework that learns a mapping between variable-length input and output sequences in one step using a purely data-driven method. The attention model has often been shown to improve the performance over another end-to-end approach, the Connectionist Temporal Classification (CTC), mainly because it explicitly uses the history of the target character without any conditional independence assumptions. However, we observed that the performance of the attention has shown poor results in noisy condition and is hard to learn in the initial training stage with long input sequences. This is because the attention model is too flexible to predict proper alignments in such cases due to the lack of left-to-right constraints as used in CTC. This paper presents a novel method for end-to-end speech recognition to improve robustness and achieve fast convergence by using a joint CTC-attention model within the multi-task learning framework, thereby mitigating the alignment issue. An experiment on the WSJ and CHiME-4 tasks demonstrates its advantages over both the CTC and attention-based encoder-decoder baselines, showing 5.4-14.6 relative improvements in Character Error Rate (CER).",
"We consider the scene text recognition problem under the attention-based encoder-decoder framework, which is the state of the art. The existing methods usually employ a frame-wise maximal likelihood loss to optimize the models. When we train the model, the misalignment between the ground truth strings and the attention's output sequences of probability distribution, which is caused by missing or superfluous characters, will confuse and mislead the training process, and consequently make the training costly and degrade the recognition accuracy. To handle this problem, we propose a novel method called edit probability (EP) for scene text recognition. EP tries to effectively estimate the probability of generating a string from the output sequence of probability distribution conditioned on the input image, while considering the possible occurrences of missing superfluous characters. The advantage lies in that the training process can focus on the missing, superfluous and unrecognized characters, and thus the impact of the misalignment problem can be alleviated or even overcome. We conduct extensive experiments on standard benchmarks, including the IIIT-5K, Street View Text and ICDAR datasets. Experimental results show that the EP can substantially boost scene text recognition performance.",
"We consider the scene text recognition problem under the attention-based encoder-decoder framework, which is the state of the art. The existing methods usually employ a frame-wise maximal likelihood loss to optimize the models. When we train the model, the misalignment between the ground truth strings and the attention's output sequences of probability distribution, which is caused by missing or superfluous characters, will confuse and mislead the training process, and consequently make the training costly and degrade the recognition accuracy. To handle this problem, we propose a novel method called edit probability (EP) for scene text recognition. EP tries to effectively estimate the probability of generating a string from the output sequence of probability distribution conditioned on the input image, while considering the possible occurrences of missing superfluous characters. The advantage lies in that the training process can focus on the missing, superfluous and unrecognized characters, and thus the impact of the misalignment problem can be alleviated or even overcome. We conduct extensive experiments on standard benchmarks, including the IIIT-5K, Street View Text and ICDAR datasets. Experimental results show that the EP can substantially boost scene text recognition performance.",
"We explore an approach to forecasting human motion in a few milliseconds given an input 3D skeleton sequence based on a recurrent encoder-decoder framework. Current approaches suffer from the problem of prediction discontinuities and may fail to predict human-like motion in longer time horizons due to error accumulation. We address these critical issues by incorporating local geometric structure constraints and regularizing predictions with plausible temporal smoothness and continuity from a global perspective. Specifically, rather than using the conventional Euclidean loss, we propose a novel frame-wise geodesic loss as a geometrically meaningful, more precise distance measurement. Moreover, inspired by the adversarial training mechanism, we present a new learning procedure to simultaneously validate the sequence-level plausibility of the prediction and its coherence with the input sequence by introducing two global recurrent discriminators. An unconditional, fidelity discriminator and a conditional, continuity discriminator are jointly trained along with the predictor in an adversarial manner. Our resulting adversarial geometry-aware encoder-decoder (AGED) model significantly outperforms state-of-the-art deep learning based approaches on the heavily benchmarked H3.6M dataset in both short-term and long-term predictions.",
"The standard content-based attention mechanism typically used in sequence-to-sequence models is computationally expensive as it requires the comparison of large encoder and decoder states at each time step. In this work, we propose an alternative attention mechanism based on a fixed size memory representation that is more efficient. Our technique predicts a compact set of K attention contexts during encoding and lets the decoder compute an efficient lookup that does not need to consult the memory. We show that our approach performs on-par with the standard attention mechanism while yielding inference speedups of 20 for real-world translation tasks and more for tasks with longer sequences. By visualizing attention scores we demonstrate that our models learn distinct, meaningful alignments.",
"Attention-based neural encoder-decoder frameworks have been widely adopted for image captioning. Most methods force visual attention to be active for every generated word. However, the decoder likely requires little to no visual information from the image to predict non-visual words such as \"the\" and \"of\". Other words that may seem visual can often be predicted reliably just from the language model e.g., \"sign\" after \"behind a red stop\" or \"phone\" following \"talking on a cell\". In this paper, we propose a novel adaptive attention model with a visual sentinel. At each time step, our model decides whether to attend to the image (and if so, to which regions) or to the visual sentinel. The model decides whether to attend to the image and where, in order to extract meaningful information for sequential word generation. We test our method on the COCO image captioning 2015 challenge dataset and Flickr30K. Our approach sets the new state-of-the-art by a significant margin.",
"This paper investigates the framework of encoder-decoder with attention for sequence labelling based spoken language understanding. We introduce Bidirectional Long Short Term Memory - Long Short Term Memory networks (BLSTM-LSTM) as the encoder-decoder model to fully utilize the power of deep learning. In the sequence labelling task, the input and output sequences are aligned word by word, while the attention mechanism cannot provide the exact alignment. To address this limitation, we propose a novel focus mechanism for encoder-decoder framework. Experiments on the standard ATIS dataset showed that BLSTM-LSTM with focus mechanism defined the new state-of-the-art by outperforming standard BLSTM and attention based encoder-decoder. Further experiments also show that the proposed model is more robust to speech recognition errors.",
"This paper investigates the framework of encoder-decoder with attention for sequence labelling based spoken language understanding. We introduce Bidirectional Long Short Term Memory - Long Short Term Memory networks (BLSTM-LSTM) as the encoder-decoder model to fully utilize the power of deep learning. In the sequence labelling task, the input and output sequences are aligned word by word, while the attention mechanism cannot provide the exact alignment. To address this limitation, we propose a novel focus mechanism for encoder-decoder framework. Experiments on the standard ATIS dataset showed that BLSTM-LSTM with focus mechanism defined the new state-of-the-art by outperforming standard BLSTM and attention based encoder-decoder. Further experiments also show that the proposed model is more robust to speech recognition errors.",
"A fast reference frame selection algorithm for the HEVC encoder is proposed.A relationship between the content similarity and reference frame selection is derived.The content similarity is studied without any extra computational complexity.Experimental results show that the proposed algorithm efficiently removes the encoding complexity of the best reference frame decision process. The high efficiency video coding (HEVC) is the state-of-the-art video coding standard, which achieves about 50 bit rate saving while maintaining the same visual quality as compared to the H.264 AVC. This achieved coding efficiency benefits from a set of advanced coding tools, such as the multiple reference frames (MRF) based interframe prediction, which efficiently improves the coding efficiency of the HEVC encoder, while it also increases heavy computation into the HEVC encoder. The high encoding complexity becomes a bottleneck for the high definition videos and HEVC encoder to be widely used in real-time and low power multimedia applications. In this paper, we propose a content similarity based fast reference frame selection algorithm for reducing the computational complexity of the multiple reference frames based interframe prediction. Based the large content similarity between the parent prediction unit (Inter_2N2N) and the children prediction units (Inter_2NN, Inter_N2N, Inter_NN, Inter_2NnU, Inter_2NnD, Inter_nL2N, and Inter_nR2N), the reference frame selection information of the children prediction units are obtained by learning the results of their parent prediction unit. Experimental results show that the proposed algorithm can reduce about 54.29 and 43.46 MRF encoding time saving for the low-delay-main and random-access-main coding structures, respectively, while the rate distortion performance degradation is negligible."
]
} |
1907.09705 | 2962957458 | Scene text recognition has been an important, active research topic in computer vision for years. Previous approaches mainly consider text as 1D signals and cast scene text recognition as a sequence prediction problem, by feat of CTC or attention based encoder-decoder framework, which is originally designed for speech recognition. However, different from speech voices, which are 1D signals, text instances are essentially distributed in 2D image spaces. To adhere to and make use of the 2D nature of text for higher recognition accuracy, we extend the vanilla CTC model to a second dimension, thus creating 2D-CTC. 2D-CTC can adaptively concentrate on most relevant features while excluding the impact from clutters and noises in the background; It can also naturally handle text instances with various forms (horizontal, oriented and curved) while giving more interpretable intermediate predictions. The experiments on standard benchmarks for scene text recognition, such as IIIT-5K, ICDAR 2015, SVP-Perspective, and CUTE80, demonstrate that the proposed 2D-CTC model outperforms state-of-the-art methods on the text of both regular and irregular shapes. Moreover, 2D-CTC exhibits its superiority over prior art on training and testing speed. Our implementation and models of 2D-CTC will be made publicly available soon later. | In contrast to the aforementioned methods, Liao al @cite_19 recently propose to utilize instance segmentation to simultaneously predict character locations and recognition results, avoiding the problem of attention misalignment. They also notice the conflict between the 2D image feature and collapsed sequence presentation, and propose a reasonable solution. However, this method requires character-level annotations, and are limited to real-world applications, especially for areas where detailed annotations are hardly available (e.g., handwritten text recognition). | {
"cite_N": [
"@cite_19"
],
"mid": [
"2751748110"
],
"abstract": [
"Scene text recognition has attracted great interests from the computer vision and pattern recognition community in recent years. State-of-the-art methods use concolutional neural networks (CNNs), recurrent neural networks with long short-term memory (RNN-LSTM) or the combination of them. In this paper, we investigate the intrinsic characteristics of text recognition, and inspired by human cognition mechanisms in reading texts, we propose a scene text recognition method with character models on convolutional feature map. The method simultaneously detects and recognizes characters by sliding the text line image with character models, which are learned end-to-end on text line images labeled with text transcripts. The character classifier outputs on the sliding windows are normalized and decoded with Connectionist Temporal Classification (CTC) based algorithm. Compared to previous methods, our method has a number of appealing properties: (1) It avoids the difficulty of character segmentation which hinders the performance of segmentation-based recognition methods; (2) The model can be trained simply and efficiently because it avoids gradient vanishing exploding in training RNN-LSTM based models; (3) It bases on character models trained free of lexicon, and can recognize unknown words. (4) The recognition process is highly parallel and enables fast recognition. Our experiments on several challenging English and Chinese benchmarks, including the IIIT-5K, SVT, ICDAR03 13 and TRW15 datasets, demonstrate that the proposed method yields superior or comparable performance to state-of-the-art methods while the model size is relatively small."
]
} |
1907.09705 | 2962957458 | Scene text recognition has been an important, active research topic in computer vision for years. Previous approaches mainly consider text as 1D signals and cast scene text recognition as a sequence prediction problem, by feat of CTC or attention based encoder-decoder framework, which is originally designed for speech recognition. However, different from speech voices, which are 1D signals, text instances are essentially distributed in 2D image spaces. To adhere to and make use of the 2D nature of text for higher recognition accuracy, we extend the vanilla CTC model to a second dimension, thus creating 2D-CTC. 2D-CTC can adaptively concentrate on most relevant features while excluding the impact from clutters and noises in the background; It can also naturally handle text instances with various forms (horizontal, oriented and curved) while giving more interpretable intermediate predictions. The experiments on standard benchmarks for scene text recognition, such as IIIT-5K, ICDAR 2015, SVP-Perspective, and CUTE80, demonstrate that the proposed 2D-CTC model outperforms state-of-the-art methods on the text of both regular and irregular shapes. Moreover, 2D-CTC exhibits its superiority over prior art on training and testing speed. Our implementation and models of 2D-CTC will be made publicly available soon later. | In concern of both accuracy and efficiency, 2D-CTC recognizes text from a 2D perspective similar to @cite_19 , but trained without any character-level annotations. By extending vanilla CTC, 2D-CTC achieves state-of-the-art performance, while retaining the high efficiency of CTC models. | {
"cite_N": [
"@cite_19"
],
"mid": [
"2594856242"
],
"abstract": [
"Most existing sequence labelling models rely on a fixed decomposition of a target sequence into a sequence of basic units. These methods suffer from two major drawbacks: 1) the set of basic units is fixed, such as the set of words, characters or phonemes in speech recognition, and 2) the decomposition of target sequences is fixed. These drawbacks usually result in sub-optimal performance of modeling sequences. In this pa- per, we extend the popular CTC loss criterion to alleviate these limitations, and propose a new loss function called Gram-CTC. While preserving the advantages of CTC, Gram-CTC automatically learns the best set of basic units (grams), as well as the most suitable decomposition of tar- get sequences. Unlike CTC, Gram-CTC allows the model to output variable number of characters at each time step, which enables the model to capture longer term dependency and improves the computational efficiency. We demonstrate that the proposed Gram-CTC improves CTC in terms of both performance and efficiency on the large vocabulary speech recognition task at multiple scales of data, and that with Gram-CTC we can outperform the state-of-the-art on a standard speech benchmark."
]
} |
1907.09656 | 2962692517 | Tactile perception is an essential ability of intelligent robots in interaction with their surrounding environments. This perception as an intermediate level acts between sensation and action and has to be defined properly to generate suitable action in response to sensed data. In this paper, we propose a feedback approach to address robot grasping task using force-torque tactile sensing. While visual perception is an essential part for gross reaching, constant utilization of this sensing modality can negatively affect the grasping process with overwhelming computation. In such case, human being utilizes tactile sensing to interact with objects. Inspired by, the proposed approach is presented and evaluated on a real robot to demonstrate the effectiveness of the suggested framework. Moreover, we utilize a deep learning framework called Deep Calibration in order to eliminate the effect of bias in the collected data from the robot sensors. | In this paper, we aim to propose an interactive tactile perception with use of force-torque sensing for a grasping task. Grasping task can be initiated by either visual or tactile perception. Although the first one is powerful in terms of object recognition, a continuous processing of visual data during interaction is not a simple task for a light robot. In real world, a human also barely utilizes vision ability in close proximity of the object to be grasped; indeed, tactile perception ability is more useful in the vicinity of object @cite_20 , @cite_21 . | {
"cite_N": [
"@cite_21",
"@cite_20"
],
"mid": [
"2149606722",
"2802840089"
],
"abstract": [
"Tactile information is valuable in determining properties of objects that are inaccessible from visual perception. In this paper, we present a tactile perception strategy that allows a mobile robot with tactile sensors in its gripper to measure a generic set of tactile features while manipulating an object. We propose a switching velocity-force controller that grasps an object safely and reveals, at the same time, its deformation properties. By gently rolling the object, the robot can extract additional information about the contents of the object. As an application, we show that a robot can use these features to distinguish the internal state of bottles and cans-purely from tactile sensing-from a small training set. The robot can distinguish open from closed bottles and cans and full ones from empty ones. We also show how the high-frequency component in tactile information can be used to detect movement inside a container, e.g., in order to detect the presence of liquid. To prove that this is a hard recognition problem, we also conducted a comparative study with 17 human test subjects. The recognition rates of the human subjects were comparable with that of the robot.",
"Can a robot grasp an unknown object without seeing it? In this paper, we present a tactile-sensing based approach to this challenging problem of grasping novel objects without prior knowledge of their location or physical properties. Our key idea is to combine touch based object localization with tactile based re-grasping. To train our learning models, we created a large-scale grasping dataset, including more than 30 RGB frames and over 2.8 million tactile samples from 7800 grasp interactions of 52 objects. To learn a representation of tactile signals, we propose an unsupervised auto-encoding scheme, which shows a significant improvement of 4-9 over prior methods on a variety of tactile perception tasks. Our system consists of two steps. First, our touch localization model sequentially 'touch-scans' the workspace and uses a particle filter to aggregate beliefs from multiple hits of the target. It outputs an estimate of the object's location, from which an initial grasp is established. Next, our re-grasping model learns to progressively improve grasps with tactile feedback based on the learned features. This network learns to estimate grasp stability and predict adjustment for the next grasp. Re-grasping thus is performed iteratively until our model identifies a stable grasp. Finally, we demonstrate extensive experimental results on grasping a large set of novel objects using tactile sensing alone. Furthermore, when applied on top of a vision-based policy, our re-grasping model significantly boosts the overall accuracy by 10.6 . We believe this is the first attempt at learning to grasp with only tactile sensing and without any prior object knowledge."
]
} |
1907.09656 | 2962692517 | Tactile perception is an essential ability of intelligent robots in interaction with their surrounding environments. This perception as an intermediate level acts between sensation and action and has to be defined properly to generate suitable action in response to sensed data. In this paper, we propose a feedback approach to address robot grasping task using force-torque tactile sensing. While visual perception is an essential part for gross reaching, constant utilization of this sensing modality can negatively affect the grasping process with overwhelming computation. In such case, human being utilizes tactile sensing to interact with objects. Inspired by, the proposed approach is presented and evaluated on a real robot to demonstrate the effectiveness of the suggested framework. Moreover, we utilize a deep learning framework called Deep Calibration in order to eliminate the effect of bias in the collected data from the robot sensors. | Grasping is an essential and complex daily activity. Through this task, humans show an intention to affect surrounding environment in a controllable manner. Humans primarily utilize a combination of control strategy and learning from repetitive experiments to anticipate grasping in different situations @cite_12 . In this regard, the properties of an object such as size, shape, and contact surface are important parameters during grasping task @cite_4 . | {
"cite_N": [
"@cite_4",
"@cite_12"
],
"mid": [
"2047217088",
"2002236162"
],
"abstract": [
"This paper is the second in a two-part series analyzing human grasping behavior during a wide range of unstructured tasks. It investigates the tasks performed during the daily work of two housekeepers and two machinists and correlates grasp type and object properties with the attributes of the tasks being performed. The task or activity is classified according to the force required, the degrees of freedom, and the functional task type. We found that 46 percent of tasks are constrained, where the manipulated object is not allowed to move in a full six degrees of freedom. Analyzing the interrelationships between the grasp, object, and task data show that the best predictors of the grasp type are object size, task constraints, and object mass. Using these attributes, the grasp type can be predicted with 47 percent accuracy. Those parameters likely make useful heuristics for grasp planning systems. The results further suggest the common sub-categorization of grasps into power, intermediate, and precision categories may not be appropriate, indicating that grasps are generally more multi-functional than previously thought. We find large and heavy objects are grasped with a power grasp, but small and lightweight objects are not necessarily grasped with precision grasps—even with grasped object size less than 2 cm and mass less than 20 g, precision grasps are only used 61 percent of the time. These results have important implications for robotic hand design and grasp planners, since it appears while power grasps are frequently used for heavy objects, they can still be quite practical for small, lightweight objects.",
"This paper is the first of a two-part series analyzing human grasping behavior during a wide range of unstructured tasks. The results help clarify overall characteristics of human hand to inform many domains, such as the design of robotic manipulators, targeting rehabilitation toward important hand functionality, and designing haptic devices for use by the hand. It investigates the properties of objects grasped by two housekeepers and two machinists during the course of almost 10,000 grasp instances and correlates the grasp types used to the properties of the object. We establish an object classification that assigns each object properties from a set of seven classes, including mass, shape and size of the grasp location, grasped dimension, rigidity, and roundness. The results showed that 55 percent of grasped objects had at least one dimension larger than 15 cm, suggesting that more than half of objects cannot physically be grasped using their largest axis. Ninety-two percent of objects had a mass of 500 g or less, implying that a high payload capacity may be unnecessary to accomplish a large subset of human grasping behavior. In terms of grasps, 96 percent of grasp locations were 7 cm or less in width, which can help to define requirements for hand rehabilitation and defines a reasonable grasp aperture size for a robotic hand. Subjects grasped the smallest overall major dimension of the object in 94 percent of the instances. This suggests that grasping the smallest axis of an object could be a reliable default behavior to implement in grasp planners."
]
} |
1907.09656 | 2962692517 | Tactile perception is an essential ability of intelligent robots in interaction with their surrounding environments. This perception as an intermediate level acts between sensation and action and has to be defined properly to generate suitable action in response to sensed data. In this paper, we propose a feedback approach to address robot grasping task using force-torque tactile sensing. While visual perception is an essential part for gross reaching, constant utilization of this sensing modality can negatively affect the grasping process with overwhelming computation. In such case, human being utilizes tactile sensing to interact with objects. Inspired by, the proposed approach is presented and evaluated on a real robot to demonstrate the effectiveness of the suggested framework. Moreover, we utilize a deep learning framework called Deep Calibration in order to eliminate the effect of bias in the collected data from the robot sensors. | Tactile sensation is a very informative feature to recognize object properties. In @cite_19 , the authors proposed a tactile perception strategy to measure tactile features for mobile robots. Tactile sensing also is used to propose a robust controller for reliable grasping @cite_17 and slipping avoidance @cite_0 . Visual sensing and tactile sensing are complementary in robot grasping. A combination of both of them through deep architecture is a promising solution in @cite_2 , @cite_6 . However, processing these high-dimensional data is not an easy task and a meaningful compact representation would be needed @cite_9 , @cite_23 . A robot can learn the manipulation using tactile sensation through demonstrations @cite_1 , @cite_9 , @cite_14 . | {
"cite_N": [
"@cite_14",
"@cite_9",
"@cite_1",
"@cite_6",
"@cite_0",
"@cite_19",
"@cite_23",
"@cite_2",
"@cite_17"
],
"mid": [
"2149606722",
"2962983231",
"2793447234",
"1560331997",
"2035557456",
"2775635818",
"2802840089",
"1979709633",
"2963263790"
],
"abstract": [
"Tactile information is valuable in determining properties of objects that are inaccessible from visual perception. In this paper, we present a tactile perception strategy that allows a mobile robot with tactile sensors in its gripper to measure a generic set of tactile features while manipulating an object. We propose a switching velocity-force controller that grasps an object safely and reveals, at the same time, its deformation properties. By gently rolling the object, the robot can extract additional information about the contents of the object. As an application, we show that a robot can use these features to distinguish the internal state of bottles and cans-purely from tactile sensing-from a small training set. The robot can distinguish open from closed bottles and cans and full ones from empty ones. We also show how the high-frequency component in tactile information can be used to detect movement inside a container, e.g., in order to detect the presence of liquid. To prove that this is a hard recognition problem, we also conducted a comparative study with 17 human test subjects. The recognition rates of the human subjects were comparable with that of the robot.",
"Vision and touch are two of the important sensing modalities for humans and they offer complementary information for sensing the environment. Robots could also benefit from such multi-modal sensing ability. In this paper, addressing for the first time (to the best of our knowledge) texture recognition from tactile images and vision, we propose a new fusion method named Deep Maximum Covariance Analysis (DMCA) to learn a joint latent space for sharing features through vision and tactile sensing. The features of camera images and tactile data acquired from a GelSight sensor are learned by deep neural networks. But the learned features are of a high dimensionality and are redundant due to the differences between the two sensing modalities, which deteriorates the perception performance. To address this, the learned features are paired using maximum covariance analysis. Results of the algorithm on a newly collected dataset of paired visual and tactile data relating to cloth textures show that a good recognition performance of greater than 90 can be achieved by using the proposed DMCA framework. In addition, we find that the perception performance of either vision or tactile sensing can be improved by employing the shared representation space, compared to learning from unimodal data.",
"Tactile sensing is required for human-like control with robotic manipulators. Multimodality is an essential component for these tactile sensors, for robots to achieve both the perceptual accuracy required for precise control, as well as the robustness to maintain a stable grasp without causing damage to the object or the robot itself. In this study, we present a cheap, 3D-printed, compliant, dual-modal, optical tactile sensor that is capable of both high (temporal) speed sensing, analogous to pain reception in humans and high (spatial) resolution sensing, analogous to the sensing provided by Merkel cell complexes in the human fingertip. We apply three tasks for testing the sensing capabilities in both modes; first, a depth modulation task, requiring the robot to follow a target trajectory using the high-speed mode; second, a high-resolution perception task, where the sensor perceives angle and radial position relative to an object edge; and third, a tactile exploration task, where the robot uses the high-resolution mode to perceive an edge and subsequently follow the object contour. The robot is capable of modulating contact depth using the high-speed mode, high accuracy in the perception task, and accurate control using the high-resolution mode.",
"This paper presents a novel framework for integration of vision and tactile sensing by localizing tactile readings in a visual object map. Intuitively, there are some correspondences, e.g., prominent features, between visual and tactile object identification. To apply it in robotics, we propose to localize tactile readings in visual images by sharing same sets of feature descriptors through two sensing modalities. It is then treated as a probabilistic estimation problem solved in a framework of recursive Bayesian filtering. Feature-based measurement model and Gaussian based motion model are thus built. In our tests, a tactile array sensor is utilized to generate tactile images during interaction with objects and the results have proven the feasibility of our proposed framework.",
"It is frequently accepted in the manipulation literature that tactile sensing is needed to improve the precision of robot manipulation. However, there is no consensus on how this may be achieved. This paper applies particle filtering to the problem of localizing the pose and shape of an object that the robot touches. We are motivated by the situation where the robot has enclosed its fingers around an object but has not yet grasped it. This might be the case just prior to grasping or when the robot is holding on to something fixtured elsewhere in the environment. In order to solve this problem, we propose a new model for position measurements of points on the robot manipulator that tactile sensing indicates are touching the object. We also propose a model for points on the manipulator that tactile measurements indicate are not touching the object. Finally, we characterize the approach in simulation and use it to localize an object that Robonaut 2 holds in its hand.",
"Tactile sensing is an important perception mode for robots, but the existing tactile technologies have multiple limitations. What kind of tactile information robots need, and how to use the information, remain open questions. We believe a soft sensor surface and high-resolution sensing of geometry should be important components of a competent tactile sensor. In this paper, we discuss the development of a vision-based optical tactile sensor, GelSight. Unlike the traditional tactile sensors which measure contact force, GelSight basically measures geometry, with very high spatial resolution. The sensor has a contact surface of soft elastomer, and it directly measures its deformation, both vertical and lateral, which corresponds to the exact object shape and the tension on the contact surface. The contact force, and slip can be inferred from the sensor’s deformation as well. Particularly, we focus on the hardware and software that support GelSight’s application on robot hands. This paper reviews the development of GelSight, with the emphasis in the sensing principle and sensor design. We introduce the design of the sensor’s optical system, the algorithm for shape, force and slip measurement, and the hardware designs and fabrication of different sensor versions. We also show the experimental evaluation on the GelSight’s performance on geometry and force measurement. With the high-resolution measurement of shape and contact force, the sensor has successfully assisted multiple robotic tasks, including material perception or recognition and in-hand localization for robot manipulation.",
"Can a robot grasp an unknown object without seeing it? In this paper, we present a tactile-sensing based approach to this challenging problem of grasping novel objects without prior knowledge of their location or physical properties. Our key idea is to combine touch based object localization with tactile based re-grasping. To train our learning models, we created a large-scale grasping dataset, including more than 30 RGB frames and over 2.8 million tactile samples from 7800 grasp interactions of 52 objects. To learn a representation of tactile signals, we propose an unsupervised auto-encoding scheme, which shows a significant improvement of 4-9 over prior methods on a variety of tactile perception tasks. Our system consists of two steps. First, our touch localization model sequentially 'touch-scans' the workspace and uses a particle filter to aggregate beliefs from multiple hits of the target. It outputs an estimate of the object's location, from which an initial grasp is established. Next, our re-grasping model learns to progressively improve grasps with tactile feedback based on the learned features. This network learns to estimate grasp stability and predict adjustment for the next grasp. Re-grasping thus is performed iteratively until our model identifies a stable grasp. Finally, we demonstrate extensive experimental results on grasping a large set of novel objects using tactile sensing alone. Furthermore, when applied on top of a vision-based policy, our re-grasping model significantly boosts the overall accuracy by 10.6 . We believe this is the first attempt at learning to grasp with only tactile sensing and without any prior object knowledge.",
"Tactile sensing is a fundamental component of object manipulation and tool handling skills. With robots entering unstructured environments, tactile feedback also becomes an important ability for robot manipulation. In this work, we explore how a robot can learn to use tactile sensing in object manipulation tasks. We first address the problem of in-hand object localization and adapt three pose estimation algorithms from computer vision. Second, we employ dynamic motor primitives to learn robot movements from human demonstrations and record desired tactile signal trajectories. Then, we add tactile feedback to the control loop and apply relative entropy policy search to learn the parameters of the tactile coupling. Additionally, we show how the learning of tactile feedback can be performed more efficiently by reducing the dimensionality of the tactile information through spectral clustering and principal component analysis. Our approach is implemented on a real robot, which learns to perform a scraping task with a spatula in an altered environment.",
"Touch sensing can help robots understand their surrounding environment, and in particular the objects they interact with. To this end, roboticists have, in the last few decades, developed several tactile sensing solutions, extensively reported in the literature. Research into interpreting the conveyed tactile information has also started to attract increasing attention in recent years. However, a comprehensive study on this topic is yet to be reported. In an effort to collect and summarize the major scientific achievements in the area, this survey extensively reviews current trends in robot tactile perception of object properties. Available tactile sensing technologies are briefly presented before an extensive review on tactile recognition of object properties. The object properties that are targeted by this review are shape, surface material and object pose. The role of touch sensing in combination with other sensing sources is also discussed. In this review, open issues are identified and future directions for applying tactile sensing in different tasks are suggested."
]
} |
1907.09656 | 2962692517 | Tactile perception is an essential ability of intelligent robots in interaction with their surrounding environments. This perception as an intermediate level acts between sensation and action and has to be defined properly to generate suitable action in response to sensed data. In this paper, we propose a feedback approach to address robot grasping task using force-torque tactile sensing. While visual perception is an essential part for gross reaching, constant utilization of this sensing modality can negatively affect the grasping process with overwhelming computation. In such case, human being utilizes tactile sensing to interact with objects. Inspired by, the proposed approach is presented and evaluated on a real robot to demonstrate the effectiveness of the suggested framework. Moreover, we utilize a deep learning framework called Deep Calibration in order to eliminate the effect of bias in the collected data from the robot sensors. | A lower dimensional representation of tactile data is also more useful for object material classification @cite_16 . With more processing approaches such as bag-of-words, identification of objects would be possible in advance @cite_22 . Extraction of object pose via touch based perception can be used for manipulation @cite_15 . Moreover, localization will be improved by contact information gathered by tactile sensor @cite_7 , @cite_5 . The robot can control and adjust the pose of hand with stability consideration after evaluating the tactile experiences @cite_18 , @cite_11 . | {
"cite_N": [
"@cite_18",
"@cite_22",
"@cite_7",
"@cite_5",
"@cite_15",
"@cite_16",
"@cite_11"
],
"mid": [
"2035557456",
"2421020186",
"2802840089",
"2149606722",
"2472910955",
"2962983231",
"2138648596"
],
"abstract": [
"It is frequently accepted in the manipulation literature that tactile sensing is needed to improve the precision of robot manipulation. However, there is no consensus on how this may be achieved. This paper applies particle filtering to the problem of localizing the pose and shape of an object that the robot touches. We are motivated by the situation where the robot has enclosed its fingers around an object but has not yet grasped it. This might be the case just prior to grasping or when the robot is holding on to something fixtured elsewhere in the environment. In order to solve this problem, we propose a new model for position measurements of points on the robot manipulator that tactile sensing indicates are touching the object. We also propose a model for points on the manipulator that tactile measurements indicate are not touching the object. Finally, we characterize the approach in simulation and use it to localize an object that Robonaut 2 holds in its hand.",
"Robust manipulation and insertion of small parts can be challenging because of the small tolerances typically involved. The key to robust control of these kinds of manipulation interactions is accurate tracking and control of the parts involved. Typically, this is accomplished using visual servoing or force-based control. However, these approaches have drawbacks. Instead, we propose a new approach that uses tactile sensing to accurately localize the pose of a part grasped in the robot hand. Using a feature-based matching technique in conjunction with a newly developed tactile sensing technology known as GelSight that has much higher resolution than competing methods, we synthesize high-resolution height maps of object surfaces. As a result of these high-resolution tactile maps, we are able to localize small parts held in a robot hand very accurately. We quantify localization accuracy in benchtop experiments and experimentally demonstrate the practicality of the approach in the context of a small parts insertion problem.",
"Can a robot grasp an unknown object without seeing it? In this paper, we present a tactile-sensing based approach to this challenging problem of grasping novel objects without prior knowledge of their location or physical properties. Our key idea is to combine touch based object localization with tactile based re-grasping. To train our learning models, we created a large-scale grasping dataset, including more than 30 RGB frames and over 2.8 million tactile samples from 7800 grasp interactions of 52 objects. To learn a representation of tactile signals, we propose an unsupervised auto-encoding scheme, which shows a significant improvement of 4-9 over prior methods on a variety of tactile perception tasks. Our system consists of two steps. First, our touch localization model sequentially 'touch-scans' the workspace and uses a particle filter to aggregate beliefs from multiple hits of the target. It outputs an estimate of the object's location, from which an initial grasp is established. Next, our re-grasping model learns to progressively improve grasps with tactile feedback based on the learned features. This network learns to estimate grasp stability and predict adjustment for the next grasp. Re-grasping thus is performed iteratively until our model identifies a stable grasp. Finally, we demonstrate extensive experimental results on grasping a large set of novel objects using tactile sensing alone. Furthermore, when applied on top of a vision-based policy, our re-grasping model significantly boosts the overall accuracy by 10.6 . We believe this is the first attempt at learning to grasp with only tactile sensing and without any prior object knowledge.",
"Tactile information is valuable in determining properties of objects that are inaccessible from visual perception. In this paper, we present a tactile perception strategy that allows a mobile robot with tactile sensors in its gripper to measure a generic set of tactile features while manipulating an object. We propose a switching velocity-force controller that grasps an object safely and reveals, at the same time, its deformation properties. By gently rolling the object, the robot can extract additional information about the contents of the object. As an application, we show that a robot can use these features to distinguish the internal state of bottles and cans-purely from tactile sensing-from a small training set. The robot can distinguish open from closed bottles and cans and full ones from empty ones. We also show how the high-frequency component in tactile information can be used to detect movement inside a container, e.g., in order to detect the presence of liquid. To prove that this is a hard recognition problem, we also conducted a comparative study with 17 human test subjects. The recognition rates of the human subjects were comparable with that of the robot.",
"This paper addresses 6-DOF (degree-of-freedom) tactile localization, i.e., the pose estimation of tridimensional objects using tactile measurements. This estimation problem is fundamental for the operation of autonomous robots that are often required to manipulate and grasp objects whose pose is a priori unknown. The nature of tactile measurements, the strict time requirements for real-time operation, and the multimodality of the involved probability distributions pose remarkable challenges and call for advanced nonlinear filtering techniques. Following a Bayesian approach, this paper proposes a novel and effective algorithm, named memory unscented particle filter (MUPF), which solves 6-DOF localization recursively in real time by only exploiting contact point measurements. The MUPF combines a modified particle filter that incorporates a sliding memory of past measurements to better handle multimodal distributions, along with the unscented Kalman filter that moves the particles toward regions of the search space that are more likely with the measurements. The performance of the proposed MUPF algorithm has been assessed both in simulation and on a real robotic system equipped with tactile sensors (i.e., the iCub humanoid robot). The experiments show that the algorithm provides accurate and reliable localization even with a low number of particles and, hence, is compatible with real-time requirements.",
"Vision and touch are two of the important sensing modalities for humans and they offer complementary information for sensing the environment. Robots could also benefit from such multi-modal sensing ability. In this paper, addressing for the first time (to the best of our knowledge) texture recognition from tactile images and vision, we propose a new fusion method named Deep Maximum Covariance Analysis (DMCA) to learn a joint latent space for sharing features through vision and tactile sensing. The features of camera images and tactile data acquired from a GelSight sensor are learned by deep neural networks. But the learned features are of a high dimensionality and are redundant due to the differences between the two sensing modalities, which deteriorates the perception performance. To address this, the learned features are paired using maximum covariance analysis. Results of the algorithm on a newly collected dataset of paired visual and tactile data relating to cloth textures show that a good recognition performance of greater than 90 can be achieved by using the proposed DMCA framework. In addition, we find that the perception performance of either vision or tactile sensing can be improved by employing the shared representation space, compared to learning from unimodal data.",
"In this paper, we present a novel approach for identifying objects using touch sensors installed in the finger tips of a manipulation robot. Our approach operates on low-resolution intensity images that are obtained when the robot grasps an object. We apply a bag-of-words approach for object identification. By means of unsupervised clustering on training data, our approach learns a vocabulary from tactile observations which is used to generate a histogram codebook. The histogram codebook models distributions over the vocabulary and is the core identification mechanism. As the objects are larger than the sensor, the robot typically needs multiple grasp actions at different positions to uniquely identify an object. To reduce the number of required grasp actions, we apply a decision-theoretic framework that minimizes the entropy of the probabilistic belief about the type of the object. In our experiments carried out with various industrial and household objects, we demonstrate that our approach is able to discriminate between a large set of objects. We furthermore show that using our approach, a robot is able to distinguish visually similar objects that have different elasticity properties by using only the information from the touch sensor."
]
} |
1907.09511 | 2963009244 | Most state-of-the-art person re-identification (re-id) methods depend on supervised model learning with a large set of cross-view identity labelled training data. Even worse, such trained models are limited to only the same-domain deployment with significantly degraded cross-domain generalization capability, i.e. "domain specific". To solve this limitation, there are a number of recent unsupervised domain adaptation and unsupervised learning methods that leverage unlabelled target domain training data. However, these methods need to train a separate model for each target domain as supervised learning methods. This conventional " train once, run once " pattern is unscalable to a large number of target domains typically encountered in real-world deployments. We address this problem by presenting a "train once, run everywhere" pattern industry-scale systems are desperate for. We formulate a "universal model learning' approach enabling domain-generic person re-id using only limited training data of a " single " seed domain. Specifically, we train a universal re-id deep model to discriminate between a set of transformed person identity classes. Each of such classes is formed by applying a variety of random appearance transformations to the images of that class, where the transformations simulate the camera viewing conditions of any domains for making the model training domain generic. Extensive evaluations show the superiority of our method for universal person re-id over a wide variety of state-of-the-art unsupervised domain adaptation and unsupervised learning re-id methods on five standard benchmarks: Market-1501, DukeMTMC, CUHK03, MSMT17, and VIPeR. | Unsupervised domain adaptation person re-id. The limitation of supervised learning re-id methods in cross-domain scalability can be addressed by using unsupervised domain adaptation (UDA) techniques. Existing UDA re-id methods generally fall into two categories: (1) image synthesis @cite_25 @cite_50 @cite_43 @cite_11 , and (2) feature alignment @cite_19 @cite_21 @cite_40 @cite_2 @cite_23 . The former aims to transfer the labelled source identity classes from the source domain to the target domain through cross-domain conditional image generation in the appearance style and background context at pixel level. The synthetic images are then used to fine-tune the model towards the target domain. On the contrary, the latter transfers the discriminative feature information learned from the labelled source training data to the target feature space by distribution alignment. These methods often use discrete attribute labels for facilitating the information transition across domains due to their better domain invariance property than low-level feature representations. | {
"cite_N": [
"@cite_21",
"@cite_19",
"@cite_43",
"@cite_40",
"@cite_50",
"@cite_2",
"@cite_23",
"@cite_25",
"@cite_11"
],
"mid": [
"2896016251",
"2963557071",
"2794651663",
"2769088658",
"2963000559",
"2901021011",
"2964055354",
"2798377719",
"2771773135"
],
"abstract": [
"Person re-identification (re-ID) poses unique challenges for unsupervised domain adaptation (UDA) in that classes in the source and target sets (domains) are entirely different and that image variations are largely caused by cameras. Given a labeled source training set and an unlabeled target training set, we aim to improve the generalization ability of re-ID models on the target testing set. To this end, we introduce a Hetero-Homogeneous Learning (HHL) method. Our method enforces two properties simultaneously: (1) camera invariance, learned via positive pairs formed by unlabeled target images and their camera style transferred counterparts; (2) domain connectedness, by regarding source target images as negative matching pairs to the target source images. The first property is implemented by homogeneous learning because training pairs are collected from the same domain. The second property is achieved by heterogeneous learning because we sample training pairs from both the source and target domains. On Market-1501, DukeMTMC-reID and CUHK03, we show that the two properties contribute indispensably and that very competitive re-ID UDA accuracy is achieved. Code is available at: https: github.com zhunzhong07 HHL.",
"Most existing person re-identification (re-id) methods require supervised model learning from a separate large set of pairwise labelled training data for every single camera pair. This significantly limits their scalability and usability in real-world large scale deployments with the need for performing re-id across many camera views. To address this scalability problem, we develop a novel deep learning method for transferring the labelled information of an existing dataset to a new unseen (unlabelled) target domain for person re-id without any supervised learning in the target domain. Specifically, we introduce an Transferable Joint Attribute-Identity Deep Learning (TJ-AIDL) for simultaneously learning an attribute-semantic and identity-discriminative feature representation space transferrable to any new (unseen) target domain for re-id tasks without the need for collecting new labelled training data from the target domain (i.e. unsupervised learning in the target domain). Extensive comparative evaluations validate the superiority of this new TJ-AIDL model for unsupervised person re-id over a wide range of state-of-the-art methods on four challenging benchmarks including VIPeR, PRID, Market-1501, and DukeMTMC-ReID.",
"Most existing person re-identification (re-id) methods require supervised model learning from a separate large set of pairwise labelled training data for every single camera pair. This significantly limits their scalability and usability in real-world large scale deployments with the need for performing re-id across many camera views. To address this scalability problem, we develop a novel deep learning method for transferring the labelled information of an existing dataset to a new unseen (unlabelled) target domain for person re-id without any supervised learning in the target domain. Specifically, we introduce an Transferable Joint Attribute-Identity Deep Learning (TJ-AIDL) for simultaneously learning an attribute-semantic and identitydiscriminative feature representation space transferrable to any new (unseen) target domain for re-id tasks without the need for collecting new labelled training data from the target domain (i.e. unsupervised learning in the target domain). Extensive comparative evaluations validate the superiority of this new TJ-AIDL model for unsupervised person re-id over a wide range of state-of-the-art methods on four challenging benchmarks including VIPeR, PRID, Market-1501, and DukeMTMC-ReID.",
"Person re-identification (re-ID) models trained on one domain often fail to generalize well to another. In our attempt, we present a \"learning via translation\" framework. In the baseline, we translate the labeled images from source to target domain in an unsupervised manner. We then train re-ID models with the translated images by supervised methods. Yet, being an essential part of this framework, unsupervised image-image translation suffers from the information loss of source-domain labels during translation. Our motivation is two-fold. First, for each image, the discriminative cues contained in its ID label should be maintained after translation. Second, given the fact that two domains have entirely different persons, a translated image should be dissimilar to any of the target IDs. To this end, we propose to preserve two types of unsupervised similarities, 1) self-similarity of an image before and after translation, and 2) domain-dissimilarity of a translated source image and a target image. Both constraints are implemented in the similarity preserving generative adversarial network (SPGAN) which consists of an Siamese network and a CycleGAN. Through domain adaptation experiment, we show that images generated by SPGAN are more suitable for domain adaptation and yield consistent and competitive re-ID accuracy on two large-scale datasets.",
"Person re-identification (re-ID) models trained on one domain often fail to generalize well to another. In our attempt, we present a \"learning via translation\" framework. In the baseline, we translate the labeled images from source to target domain in an unsupervised manner. We then train re-ID models with the translated images by supervised methods. Yet, being an essential part of this framework, unsupervised image-image translation suffers from the information loss of source-domain labels during translation. Our motivation is two-fold. First, for each image, the discriminative cues contained in its ID label should be maintained after translation. Second, given the fact that two domains have entirely different persons, a translated image should be dissimilar to any of the target IDs. To this end, we propose to preserve two types of unsupervised similarities, 1) self-similarity of an image before and after translation, and 2) domain-dissimilarity of a translated source image and a target image. Both constraints are implemented in the similarity preserving generative adversarial network (SPGAN) which consists of an Siamese network and a CycleGAN. Through domain adaptation experiment, we show that images generated by SPGAN are more suitable for domain adaptation and yield consistent and competitive re-ID accuracy on two large-scale datasets.",
"Unsupervised domain adaptation aims to mitigate the domain shift when transferring knowledge from a supervised source domain to an unsupervised target domain. Adversarial Feature Alignment has been successfully explored to minimize the domain discrepancy. However, existing methods are usually struggling to optimize mixed learning objectives and vulnerable to negative transfer when two domains do not share the identical label space. In this paper, we empirically reveal that the erratic discrimination of target domain mainly reflects in its much lower feature norm value with respect to that of the source domain. We present a non-parametric Adaptive Feature Norm AFN approach, which is independent of the association between label spaces of the two domains. We demonstrate that adapting feature norms of source and target domains to achieve equilibrium over a large range of values can result in significant domain transfer gains. Without bells and whistles but a few lines of code, our method largely lifts the discrimination of target domain (23.7 from the Source Only in VisDA2017) and achieves the new state of the art under the vanilla setting. Furthermore, as our approach does not require to deliberately align the feature distributions, it is robust to negative transfer and can outperform the existing approaches under the partial setting by an extremely large margin (9.8 on Office-Home and 14.1 on VisDA2017). Code is available at this https URL. We are responsible for the reproducibility of our method.",
"In this paper, we make two contributions to unsupervised domain adaptation (UDA) using the convolutional neural network (CNN). First, our approach transfers knowledge in all the convolutional layers through attention alignment. Most previous methods align high-level representations, e.g., activations of the fully connected (FC) layers. In these methods, however, the convolutional layers which underpin critical low-level domain knowledge cannot be updated directly towards reducing domain discrepancy. Specifically, we assume that the discriminative regions in an image are relatively invariant to image style changes. Based on this assumption, we propose an attention alignment scheme on all the target convolutional layers to uncover the knowledge shared by the source domain. Second, we estimate the posterior label distribution of the unlabeled data for target network training. Previous methods, which iteratively update the pseudo labels by the target network and refine the target network by the updated pseudo labels, are vulnerable to label estimation errors. Instead, our approach uses category distribution to calculate the cross-entropy loss for training, thereby ameliorating the error accumulation of the estimated labels. The two contributions allow our approach to outperform the state-of-the-art methods by +2.6 on the Office-31 dataset.",
"Unsupervised Domain Adaptation (UDA) aims to transfer domain knowledge from existing well-defined tasks to new ones where labels are unavailable. In the real-world applications, as the domain (task) discrepancies are usually uncontrollable, it is significantly motivated to match the feature distributions even if the domain discrepancies are disparate. Additionally, as no label is available in the target domain, how to successfully adapt the classifier from the source to the target domain still remains an open question. In this paper, we propose the Re-weighted Adversarial Adaptation Network (RAAN) to reduce the feature distribution divergence and adapt the classifier when domain discrepancies are disparate. Specifically, to alleviate the need of common supports in matching the feature distribution, we choose to minimize optimal transport (OT) based Earth-Mover (EM) distance and reformulate it to a minimax objective function. Utilizing this, RAAN can be trained in an end-to-end and adversarial manner. To further adapt the classifier, we propose to match the label distribution and embed it into the adversarial training. Finally, after extensive evaluation of our method using UDA datasets of varying difficulty, RAAN achieved the state-of-the-art results and outperformed other methods by a large margin when the domain shifts are disparate.",
"In this work, we present a method for unsupervised domain adaptation (UDA), where we aim to transfer knowledge from a label-rich domain (i.e., a source domain) to an unlabeled domain (i.e., a target domain). Many adversarial learning methods have been proposed for this task. These methods train domain classifier networks (i.e., a discriminator) to discriminate distinguish the features as either a source or target and train a feature generator network to mimic the discriminator.However, the domain classifier only tries to distinguish the features as a source or target and thus does not consider task-specific decision boundaries between classes. Therefore, a trained generator can generate ambiguous features near class boundaries. To solve the problem, we propose a new approach that attempts to align distributions of source and target by utilizing the task-specific decision boundaries. We propose to utilize task-specific classifiers as discriminators that try to detect target samples that are far from the support of the source. A feature generator learns to generate target features inside the support to fool the classifiers. Since the generator uses feedback from task-specific classifiers, it avoids generating target features near class boundaries. Our method outperforms other methods on several datasets of image classification and semantic segmentation."
]
} |
1907.09607 | 2964204277 | The success of deep learning in medical imaging is mostly achieved at the cost of a large labeled data set. Semi-supervised learning (SSL) provides a promising solution by leveraging the structure of unlabeled data to improve learning from a small set of labeled data. Self-ensembling is a simple approach used in SSL to encourage consensus among ensemble predictions of unknown labels, improving generalization of the model by making it more insensitive to the latent space. Currently, such an ensemble is obtained by randomization such as dropout regularization and random data augmentation. In this work, we hypothesize -- from the generalization perspective -- that self-ensembling can be improved by exploiting the stochasticity of a disentangled latent space. To this end, we present a stacked SSL model that utilizes unsupervised disentangled representation learning as the stochastic embedding for self-ensembling. We evaluate the presented model for multi-label classification using chest X-ray images, demonstrating its improved performance over related SSL models as well as the interpretability of its disentangled representations. | This work is mostly related to two lines of research: 1) SSL based on regularization with random transformations, and 2) disentangled representation learning and its use in SSL. In the former, consistency-based regularization is applied on ensemble predictions obtained by randomization techniques such as random data augmentation, dropout, and random max-pooling @cite_8 . This randomization was empirically shown to improve the generalization and stability of the SSL model, while its theoretical basis was recently shown to be related to the reduction of model sensitivity to the latent space @cite_9 @cite_10 . Motivated by this theory, in this work, we attempt to utilize the knowledge about the stochastic latent space -- obtained in unsupervised learning -- in this randomization process. | {
"cite_N": [
"@cite_9",
"@cite_10",
"@cite_8"
],
"mid": [
"2909986471",
"2583938035",
"2902804655"
],
"abstract": [
"The recently proposed semi-supervised learning methods exploit consistency loss between different predictions under random perturbations. Typically, a student model is trained to predict consistently with the targets generated by a noisy teacher. However, they ignore the fact that not all training data provide meaningful and reliable information in terms of consistency. For misclassified data, blindly minimizing the consistency loss around them can hinder learning. In this paper, we propose a novel certainty-driven consistency loss (CCL) to dynamically select data samples that have relatively low uncertainty. Specifically, we measure the variance or entropy of multiple predictions under random augmentations and dropout as an estimation of uncertainty. Then, we introduce two approaches, i.e. Filtering CCL and Temperature CCL to guide the student learn more meaningful and certain reliable targets, and hence improve the quality of the gradients backpropagated to the student. Experiments demonstrate the advantages of the proposed method over the state-of-the-art semi-supervised deep learning methods on three benchmark datasets: SVHN, CIFAR10, and CIFAR100. Our method also shows robustness to noisy labels.",
"Deep convolutional networks have achieved successful performance in data mining field. However, training large networks still remains a challenge, as the training data may be insufficient and the model can easily get overfitted. Hence the training process is usually combined with a model regularization. Typical regularizers include weight decay, Dropout, etc. In this paper, we propose a novel regularizer, named Structured Decorrelation Constraint (SDC), which is applied to the activations of the hidden layers to prevent overfitting and achieve better generalization. SDC impels the network to learn structured representations by grouping the hidden units and encouraging the units within the same group to have strong connections during the training procedure. Meanwhile, it forces the units in different groups to learn non-redundant representations by minimizing the cross-covariance between them. Compared with Dropout, SDC reduces the co-adaptions between the hidden units in an explicit way. Besides, we propose a novel approach called Reg-Conv that can help SDC to regularize the complex convolutional layers. Experiments on extensive datasets show that SDC significantly reduces overfitting and yields very meaningful improvements on classification performance (on CIFAR-10 6.22 accuracy promotion and on CIFAR-100 9.63 promotion).",
"We address the issue of learning from synthetic domain randomized data effectively. While previous works have showcased domain randomization as an effective learning approach, it lacks in challenging the learner and wastes valuable compute on generating easy examples. This can be attributed to uniform randomization over the rendering parameter distribution. In this work, firstly we provide a theoretical perspective on characteristics of domain randomization and analyze its limitations. As a solution to these limitations, we propose a novel algorithm which closes the loop between the synthetic generative model and the learner in an adversarial fashion. Our framework easily extends to the scenario when there is unlabelled target data available, thus incorporating domain adaptation. We evaluate our method on diverse vision tasks using state-of-the-art simulators for public datasets like CLEVR, Syn2Real, and VIRAT, where we demonstrate that a learner trained using adversarial data generation performs better than using a random data generation strategy."
]
} |
1907.09607 | 2964204277 | The success of deep learning in medical imaging is mostly achieved at the cost of a large labeled data set. Semi-supervised learning (SSL) provides a promising solution by leveraging the structure of unlabeled data to improve learning from a small set of labeled data. Self-ensembling is a simple approach used in SSL to encourage consensus among ensemble predictions of unknown labels, improving generalization of the model by making it more insensitive to the latent space. Currently, such an ensemble is obtained by randomization such as dropout regularization and random data augmentation. In this work, we hypothesize -- from the generalization perspective -- that self-ensembling can be improved by exploiting the stochasticity of a disentangled latent space. To this end, we present a stacked SSL model that utilizes unsupervised disentangled representation learning as the stochastic embedding for self-ensembling. We evaluate the presented model for multi-label classification using chest X-ray images, demonstrating its improved performance over related SSL models as well as the interpretability of its disentangled representations. | Besideds the approaches discussed above, there is also an active line of research in GAN-based SSL methods @cite_7 @cite_5 . The general idea is to add a classification objective to the original mini-max game and increases the capacity of the discriminator to associate the inputs to the corresponding labels. The presented work differs from this line of research by the emphasis on obtaining, regularizing, and interpreting the latent representations in SSL. | {
"cite_N": [
"@cite_5",
"@cite_7"
],
"mid": [
"2596763562",
"2964218010"
],
"abstract": [
"Generative Adversarial Nets (GANs) have shown promise in image generation and semi-supervised learning (SSL). However, existing GANs in SSL have two problems: (1) the generator and the discriminator (i.e. the classifier) may not be optimal at the same time; and (2) the generator cannot control the semantics of the generated samples. The problems essentially arise from the two-player formulation, where a single discriminator shares incompatible roles of identifying fake samples and predicting labels and it only estimates the data without considering the labels. To address the problems, we present triple generative adversarial net (Triple-GAN), which consists of three players---a generator, a discriminator and a classifier. The generator and the classifier characterize the conditional distributions between images and labels, and the discriminator solely focuses on identifying fake image-label pairs. We design compatible utilities to ensure that the distributions characterized by the classifier and the generator both converge to the data distribution. Our results on various datasets demonstrate that Triple-GAN as a unified model can simultaneously (1) achieve the state-of-the-art classification results among deep generative models, and (2) disentangle the classes and styles of the input and transfer smoothly in the data space via interpolation in the latent space class-conditionally.",
"Generative Adversarial Nets (GANs) have shown promise in image generation and semi-supervised learning (SSL). However, existing GANs in SSL have two problems: (1) the generator and the discriminator (i.e. the classifier) may not be optimal at the same time; and (2) the generator cannot control the semantics of the generated samples. The problems essentially arise from the two-player formulation, where a single discriminator shares incompatible roles of identifying fake samples and predicting labels and it only estimates the data without considering the labels. To address the problems, we present triple generative adversarial net (Triple-GAN), which consists of three players---a generator, a discriminator and a classifier. The generator and the classifier characterize the conditional distributions between images and labels, and the discriminator solely focuses on identifying fake image-label pairs. We design compatible utilities to ensure that the distributions characterized by the classifier and the generator both converge to the data distribution. Our results on various datasets demonstrate that Triple-GAN as a unified model can simultaneously (1) achieve the state-of-the-art classification results among deep generative models, and (2) disentangle the classes and styles of the input and transfer smoothly in the data space via interpolation in the latent space class-conditionally."
]
} |
1907.09607 | 2964204277 | The success of deep learning in medical imaging is mostly achieved at the cost of a large labeled data set. Semi-supervised learning (SSL) provides a promising solution by leveraging the structure of unlabeled data to improve learning from a small set of labeled data. Self-ensembling is a simple approach used in SSL to encourage consensus among ensemble predictions of unknown labels, improving generalization of the model by making it more insensitive to the latent space. Currently, such an ensemble is obtained by randomization such as dropout regularization and random data augmentation. In this work, we hypothesize -- from the generalization perspective -- that self-ensembling can be improved by exploiting the stochasticity of a disentangled latent space. To this end, we present a stacked SSL model that utilizes unsupervised disentangled representation learning as the stochastic embedding for self-ensembling. We evaluate the presented model for multi-label classification using chest X-ray images, demonstrating its improved performance over related SSL models as well as the interpretability of its disentangled representations. | An increased interest in SSL has also been seen in medical image anlaysis. The use of an unsupervised representation learning for better generalization has been investigated for the task of myocardial segmentation @cite_1 . In @cite_5 , SSL was used in a similar X-ray data set, although the scope was limited to binary classifications between normal and abnormal categories. To our knowledge, we present the first multi-label SSL that investigates disentangled learning and self-ensembling of stochastic latent space in medical image classification. | {
"cite_N": [
"@cite_5",
"@cite_1"
],
"mid": [
"2905563945",
"2431732335"
],
"abstract": [
"The availability of large-scale annotated data and the uneven separability of different data categories have become two major impediments of deep learning for image classification. In this paper, we present a semi-supervised hierarchical convolutional neural network (SS-HCNN) to address these two challenges. A large-scale unsupervised maximum margin clustering technique is designed, which splits images into a number of hierarchical clusters iteratively to learn cluster-level CNNs at parent nodes and category-level CNNs at leaf nodes. The splitting uses the similarity of CNN features to group visually similar images into the same cluster, which relieves the uneven data separability constraint. With the hierarchical cluster-level CNNs capturing certain high-level image category information, the category-level CNNs can be trained with a small amount of labeled images, and this relieves the data annotation constraint. A novel cluster splitting criterion is also designed, which automatically terminates the image clustering in the tree hierarchy. The proposed SS-HCNN has been evaluated on the CIFAR-100 and ImageNet classification datasets. The experiments show that the SS-HCNN trained using a portion of labeled training images can achieve comparable performance with other fully trained CNNs using all labeled images. Additionally, the SS-HCNN trained using all labeled images clearly outperforms other fully trained CNNs.",
"We work towards efficient methods of categorizing visual content in medical images as a precursor step to segmentation and anatomy recognition. In this paper, we address the problem of automatic detection of level position for a given cardiac CT slice. Specifically, we divide the body area depicted in chest CT into nine semantic categories each representing an area most relevant to the study of a disease and or key anatomic cardiovascular feature. Using a set of handcrafted image features together with features derived form a deep convolutional neural network (CNN), we build a classification scheme to map a given CT slice to the relevant level. Each feature group is used to train a separate support vector machine classifier. The resulting labels are then combined in a linear model, also learned from training data. We report margin zero and margin one accuracy of 91.7 and 98.8 and show that this hybrid approach is a very effective methodology for assigning a given CT image to a relatively narrow anatomic window."
]
} |
1907.09591 | 2964224524 | Modern trajectory optimization based approaches to motion planning are fast, easy to implement, and effective on a wide range of robotics tasks. However, trajectory optimization algorithms have parameters that are typically set in advance (and rarely discussed in detail). Setting these parameters properly can have a significant impact on the practical performance of the algorithm, sometimes making the difference between finding a feasible plan or failing at the task entirely. We propose a method for leveraging past experience to learn how to automatically adapt the parameters of Gaussian Process Motion Planning (GPMP) algorithms. Specifically, we propose a differentiable extension to the GPMP2 algorithm, so that it can be trained end-to-end from data. We perform several experiments that validate our algorithm and illustrate the benefits of our proposed learning-based approach to motion planning. | Recent work in structured learning techniques offer avenues towards contending with these challenges. Several methods have focused on incorporating optimization within neural network architectures. For example, @cite_8 implicitly learns to perform nonlinear least squares optimization by learning an RNN that predicts its update steps, @cite_16 learns to perform gradient descent, and @cite_18 utilizes a ODE solver within its network. Other methods like @cite_6 learn a sequential quadratic program as a layer in its network, which was later extended to solve model predictive control @cite_10 . @cite_9 learns structured dynamics models for reactive visuomotor control. Taking inspiration from this body of work, in this paper we present a differentiable inference-based motion planning technique that through its structure allows us to combine the strengths of both traditional model-based methods and modern learning methods, while mitigating their respective weaknesses. | {
"cite_N": [
"@cite_18",
"@cite_8",
"@cite_9",
"@cite_6",
"@cite_16",
"@cite_10"
],
"mid": [
"2060329955",
"2153378748",
"2762473905",
"2123940811",
"2734271713",
"2168894214"
],
"abstract": [
"Currently, the field of sensory-motor neuroscience lacks a computational model that can replicate real-time control of biological brain. Due to incomplete neural and anatomical data, traditional neural network training methods fail to model the sensory-motor systems. Here we introduce a novel modeling method based on stochastic optimal control framework which is well suited for this purpose. Our controller is implemented with a recurrent neural network (RNN) whose goal is approximating the optimal global control law for the given plant and cost function. We employ a risk-sensitive objective function proposed by Jacobson (1973) for robustness of controller. For maximum optimization efficiency, we introduce a step response sampling method, which minimizes complexity of the optimization problem. We use conjugate gradient descent method for optimization, and gradient is calculated via Pontryagins maximum principle. In the end, we obtain highly stable and robust RNN controllers that can generate infinite varieties of attractor dynamics of the plant, which are proposed as building blocks of movement generation. We show two such examples, a point attractor based and a limit-cycle based dynamics.",
"Feedforward multilayer networks trained by supervised learning have recently demonstrated state of the art performance on image labeling problems such as boundary prediction and scene parsing. As even very low error rates can limit practical usage of such systems, methods that perform closer to human accuracy remain desirable. In this work, we propose a new type of network with the following properties that address what we hypothesize to be limiting aspects of existing methods: (1) a wide' structure with thousands of features, (2) a large field of view, (3) recursive iterations that exploit statistical dependencies in label space, and (4) a parallelizable architecture that can be trained in a fraction of the time compared to benchmark multilayer convolutional networks. For the specific image labeling problem of boundary prediction, we also introduce a novel example weighting algorithm that improves segmentation accuracy. Experiments in the challenging domain of connectomic reconstruction of neural circuity from 3d electron microscopy data show that these \"Deep And Wide Multiscale Recursive\" (DAWMR) networks lead to new levels of image labeling performance. The highest performing architecture has twelve layers, interwoven supervised and unsupervised stages, and uses an input field of view of 157,464 voxels ( @math ) to make a prediction at each image location. We present an associated open source software package that enables the simple and flexible creation of DAWMR networks.",
"Optimizing deep neural networks (DNNs) often suffers from the ill-conditioned problem. We observe that the scaling-based weight space symmetry property in rectified nonlinear network will cause this negative effect. Therefore, we propose to constrain the incoming weights of each neuron to be unit-norm, which is formulated as an optimization problem over Oblique manifold. A simple yet efficient method referred to as projection based weight normalization (PBWN) is also developed to solve this problem. PBWN executes standard gradient updates, followed by projecting the updated weight back to Oblique manifold. This proposed method has the property of regularization and collaborates well with the commonly used batch normalization technique. We conduct comprehensive experiments on several widely-used image datasets including CIFAR-10, CIFAR-100, SVHN and ImageNet for supervised learning over the state-of-the-art convolutional neural networks, such as Inception, VGG and residual networks. The results show that our method is able to improve the performance of DNNs with different architectures consistently. We also apply our method to Ladder network for semi-supervised learning on permutation invariant MNIST dataset, and our method outperforms the state-of-the-art methods: we obtain test errors as 2.52 , 1.06 , and 0.91 with only 20, 50, and 100 labeled samples, respectively.",
"In this paper, we address the problem of incrementally optimizing constraint networks for maximum likelihood map learning. Our approach allows a robot to efficiently compute configurations of the network with small errors while the robot moves through the environment. We apply a variant of stochastic gradient descent and use a tree-based parameterization of the nodes in the network. By integrating adaptive learning rates in the parameterization of the network, our algorithm can use previously computed solutions to determine the result of the next optimization run. Additionally, our approach updates only the parts of the network which are affected by the newly incorporated measurements and starts the optimization approach only if the new data reveals inconsistencies with the network constructed so far. These improvements yield an efficient solution for this class of online optimization problems. Our approach has been implemented and tested on simulated and on real data. We present comparisons to recently proposed online and offline methods that address the problem of optimizing constraint network. Experiments illustrate that our approach converges faster to a network configuration with small errors than the previous approaches.",
"Deep convolutional neural networks have liberated its extraordinary power on various tasks. However, it is still very challenging to deploy state-of-the-art models into real-world applications due to their high computational complexity. How can we design a compact and effective network without massive experiments and expert knowledge? In this paper, we propose a simple and effective framework to learn and prune deep models in an end-to-end manner. In our framework, a new type of parameter -- scaling factor is first introduced to scale the outputs of specific structures, such as neurons, groups or residual blocks. Then we add sparsity regularizations on these factors, and solve this optimization problem by a modified stochastic Accelerated Proximal Gradient (APG) method. By forcing some of the factors to zero, we can safely remove the corresponding structures, thus prune the unimportant parts of a CNN. Comparing with other structure selection methods that may need thousands of trials or iterative fine-tuning, our method is trained fully end-to-end in one training pass without bells and whistles. We evaluate our method, Sparse Structure Selection with several state-of-the-art CNNs, and demonstrate very promising results with adaptive depth and width selection.",
"Our proposed deeply-supervised nets (DSN) method simultaneously minimizes classification error while making the learning process of hidden layers direct and transparent. We make an attempt to boost the classification performance by studying a new formulation in deep networks. Three aspects in convolutional neural networks (CNN) style architectures are being looked at: (1) transparency of the intermediate layers to the overall classification; (2) discriminativeness and robustness of learned features, especially in the early layers; (3) effectiveness in training due to the presence of the exploding and vanishing gradients. We introduce \"companion objective\" to the individual hidden layers, in addition to the overall objective at the output layer (a different strategy to layer-wise pre-training). We extend techniques from stochastic gradient methods to analyze our algorithm. The advantage of our method is evident and our experimental result on benchmark datasets shows significant performance gain over existing methods (e.g. all state-of-the-art results on MNIST, CIFAR-10, CIFAR-100, and SVHN)."
]
} |
1907.09133 | 2963230693 | Sensors producing 3D point clouds such as 3D laser scanners and RGB-D cameras are widely used in robotics, be it for autonomous driving or manipulation. Aligning point clouds produced by these sensors is a vital component in such applications to perform tasks such as model registration, pose estimation, and SLAM. Iterative closest point (ICP) is the most widely used method for this task, due to its simplicity and efficiency. In this paper we propose a novel method which solves the optimisation problem posed by ICP using stochastic gradient descent (SGD). Using SGD allows us to improve the convergence speed of ICP without sacrificing solution quality. Experiments using Kinect as well as Velodyne data show that, our proposed method is faster than existing methods, while obtaining solutions comparable to standard ICP. An additional benefit is robustness to parameters when processing data from different sensors. | Selecting corresponding points in the source and reference point clouds is an important step in ICP, as evidenced by the various ICP variants that have been proposed in the literature to improve this part of ICP. @cite_7 use simulated annealing to obtain a good initial starting point. Masuda and Yokoyo @cite_16 combine a least median square estimator and ICP using random samples of points in each iteration. While this method uses random sub-sampling, the goal is to remove outliers from the point cloud processed by ICP and does not perform incremental updates of the transformation estimate as our proposed method does. | {
"cite_N": [
"@cite_16",
"@cite_7"
],
"mid": [
"2561423101",
"2098764590"
],
"abstract": [
"Despite the fact that original Iterative Closest Point(ICP) algorithm has been widely used for registration, itcannot tackle the problem when two point clouds are par-tially overlapping. Accordingly, this paper proposes a ro-bust approach for the registration of partially overlappingpoint clouds. Given two initially posed clouds, it firstlybuilds up bilateral correspondence and computes bidirec-tional distances for each point in the data shape. Based onthe ratio of bidirectional distances, the exponential functionis selected and utilized to calculate the probability value,which can indicate whether the point pair belongs to theoverlapping part or not. Subsequently, the probability val-ue can be embedded into the least square function for reg-istration of partially overlapping point clouds and a novelvariant of ICP algorithm is presented to obtain the optimalrigid transformation. The proposed approach can achievegood registration of point clouds, even when their overlappercentage is low. Experimental results tested on public da-ta sets illustrate its superiority over previous approaches onrobustness.",
"In this paper we investigate the usage of persistent point feature histograms for the problem of aligning point cloud data views into a consistent global model. Given a collection of noisy point clouds, our algorithm estimates a set of robust 16D features which describe the geometry of each point locally. By analyzing the persistence of the features at different scales, we extract an optimal set which best characterizes a given point cloud. The resulted persistent features are used in an initial alignment algorithm to estimate a rigid transformation that approximately registers the input datasets. The algorithm provides good starting points for iterative registration algorithms such as ICP (Iterative Closest Point), by transforming the datasets to its convergence basin. We show that our approach is invariant to pose and sampling density, and can cope well with noisy data coming from both indoor and outdoor laser scans."
]
} |
1907.09133 | 2963230693 | Sensors producing 3D point clouds such as 3D laser scanners and RGB-D cameras are widely used in robotics, be it for autonomous driving or manipulation. Aligning point clouds produced by these sensors is a vital component in such applications to perform tasks such as model registration, pose estimation, and SLAM. Iterative closest point (ICP) is the most widely used method for this task, due to its simplicity and efficiency. In this paper we propose a novel method which solves the optimisation problem posed by ICP using stochastic gradient descent (SGD). Using SGD allows us to improve the convergence speed of ICP without sacrificing solution quality. Experiments using Kinect as well as Velodyne data show that, our proposed method is faster than existing methods, while obtaining solutions comparable to standard ICP. An additional benefit is robustness to parameters when processing data from different sensors. | Chen and Medioni @cite_2 propose a robust version of ICP which minimises the distance between points and planes (point-to-plane ICP) instead of the traditional point-to-point distance. In point-to-plane ICP, matching points in the two point clouds are determined by intersecting a ray from the source point in direction of source point's normal with the reference point's surface. This method is more robust to local minima when compared to standard ICP @cite_20 . @cite_1 combine point-to-point and point-to-plane ICP into a single probabilistic framework called Generalised-ICP (GICP). GICP is a plane-to-plane algorithm that models the local planar structure in both source and reference point clouds instead of just the reference's, as is typically done with the point-to-plane ICP. Comprehensive summaries of different ICP variants and their performance and properties can be found in @cite_18 and @cite_8 . | {
"cite_N": [
"@cite_18",
"@cite_8",
"@cite_1",
"@cite_2",
"@cite_20"
],
"mid": [
"2132512702",
"2561423101",
"2414309118",
"2140711847",
"2035997599"
],
"abstract": [
"We present a robot-pose-registration algorithm, which is entirely based on large planar-surface patches extracted from point clouds sampled from a three-dimensional (3-D) sensor. This approach offers an alternative to the traditional point-to-point iterative-closest-point (ICP) algorithm, its point-to-plane variant, as well as newer grid-based algorithms, such as the 3-D normal distribution transform (NDT). The simpler case of known plane correspondences is tackled first by deriving expressions for least-squares pose estimation considering plane-parameter uncertainty computed during plane extraction. Closed-form expressions for covariances are also derived. To round-off the solution, we present a new algorithm, which is called minimally uncertain maximal consensus (MUMC), to determine the unknown plane correspondences by maximizing geometric consistency by minimizing the uncertainty volume in configuration space. Experimental results from three 3-D sensors, viz., Swiss-Ranger, University of South Florida Odetics Laser Detection and Ranging, and an actuated SICK S300, are given. The first two have low fields of view (FOV) and moderate ranges, while the third has a much bigger FOV and range. Experimental results show that this approach is not only more robust than point- or grid-based approaches in plane-rich environments, but it is also faster, requires significantly less memory, and offers a less-cluttered planar-patches-based visualization.",
"Despite the fact that original Iterative Closest Point(ICP) algorithm has been widely used for registration, itcannot tackle the problem when two point clouds are par-tially overlapping. Accordingly, this paper proposes a ro-bust approach for the registration of partially overlappingpoint clouds. Given two initially posed clouds, it firstlybuilds up bilateral correspondence and computes bidirec-tional distances for each point in the data shape. Based onthe ratio of bidirectional distances, the exponential functionis selected and utilized to calculate the probability value,which can indicate whether the point pair belongs to theoverlapping part or not. Subsequently, the probability val-ue can be embedded into the least square function for reg-istration of partially overlapping point clouds and a novelvariant of ICP algorithm is presented to obtain the optimalrigid transformation. The proposed approach can achievegood registration of point clouds, even when their overlappercentage is low. Experimental results tested on public da-ta sets illustrate its superiority over previous approaches onrobustness.",
"We present a novel way of odometry estimation from Velodyne LiDAR point cloud scans. The aim of our work is to overcome the most painful issues of Velodyne data - the sparsity and the quantity of data points - in an efficient way, enabling more precise registration. Alignment of the point clouds which yields the final odometry is based on random sampling of the clouds using Collar Line Segments (CLS). The closest line segment pairs are identified in two sets of line segments obtained from two consequent Velodyne scans. From each pair of correspondences, a transformation aligning the matched line segments into a 3D plane is estimated. By this, significant planes (ground, walls, …) are preserved among aligned point clouds. Evaluation using the KITTI dataset shows that our method outperforms publicly available and commonly used state-of-the-art method GICP for point cloud registration in both accuracy and speed, especially in cases where the scene lacks significant landmarks or in typical urban elements. For such environments, the registration error of our method is reduced by 75 compared to the original GICP error.",
"The problem of geometric alignment of two roughly preregistered, partially overlapping, rigid, noisy 3D point sets is considered. A new natural and simple, robustified extension of the popular Iterative Closest Point (ICP) algorithm (Besl and McKay, 1992) is presented, called the Trimmed ICP (TrICP). The new algorithm is based on the consistent use of the least trimmed squares (LTS) approach in all phases of the operation. Convergence is proved and an efficient implementation is discussed. TrICP is fast, applicable to overlaps under 50 , robust to erroneous measurements and shape defects, and has easy-to-set parameters. ICP is a special case of TrICP when the overlap parameter is 100 . Results of testing the new algorithm are shown.",
"In order to solve multi-view point clouds registration in large non-feature marked scenes, a new registration and stitching method was proposed based on 2D SIFT(Scale-invariant feature transform) features. Firstly, we used texture mapping method to generate 2D effective texture image, and then extracted and match SIFT features, obtained accurate key points and registration relationship between the effective texture images. Secondly, we reflected the SIFT key points and registration relationship to the 3D point clouds data, obtained key points and registration relationship of multi-view point clouds, we can achieve multi-view point clouds stitching. Our algorithm used texture mapping method to generate 2D effective texture image, it can eliminate interference of the holes and ineffective points, and can eliminate unnecessary mistake matching. Our algorithm used correct extracted matching point pairs to stitch, avoiding stepwise iterated of ICP algorithm, so our algorithm is simple to calculate, and it's matching precision and matching efficiency are improved to some extent. We carried experiments on two-view point clouds in two large indoor; the experiment results verified the validity of our algorithm."
]
} |
1907.09133 | 2963230693 | Sensors producing 3D point clouds such as 3D laser scanners and RGB-D cameras are widely used in robotics, be it for autonomous driving or manipulation. Aligning point clouds produced by these sensors is a vital component in such applications to perform tasks such as model registration, pose estimation, and SLAM. Iterative closest point (ICP) is the most widely used method for this task, due to its simplicity and efficiency. In this paper we propose a novel method which solves the optimisation problem posed by ICP using stochastic gradient descent (SGD). Using SGD allows us to improve the convergence speed of ICP without sacrificing solution quality. Experiments using Kinect as well as Velodyne data show that, our proposed method is faster than existing methods, while obtaining solutions comparable to standard ICP. An additional benefit is robustness to parameters when processing data from different sensors. | Once the corresponding point pairs are selected the next challenge is to use this information to find the optimal transformation. Several optimisation techniques are used to minimise ICP's cost function. Fitzgibbon @cite_3 presents a method which uses a non-linear Levenberg-Marquardt method @cite_6 which combines batch gradient descent and Gauss-Newton to optimise the cost function. Levine @cite_5 uses simulated annealing to minimise the cost function. | {
"cite_N": [
"@cite_5",
"@cite_6",
"@cite_3"
],
"mid": [
"2168722300",
"1538216801",
"2049981393"
],
"abstract": [
"We show how to extend the ICP framework to nonrigid registration, while retaining the convergence properties of the original algorithm. The resulting optimal step nonrigid ICP framework allows the use of different regularisations, as long as they have an adjustable stiffness parameter. The registration loops over a series of decreasing stiffness weights, and incrementally deforms the template towards the target, recovering the whole range of global and local deformations. To find the optimal deformation for a given stiffness, optimal iterative closest point steps are used. Preliminary correspondences are estimated by a nearest-point search. Then the optimal deformation of the template for these fixed correspondences and the active stiffness is calculated. Afterwards the process continues with new correspondences found by searching from the displaced template vertices. We present an algorithm using a locally affine regularisation which assigns an affine transformation to each vertex and minimises the difference in the transformation of neighbouring vertices. It is shown that for this regularisation the optimal deformation for fixed correspondences and fixed stiffness can be determined exactly and efficiently. The method succeeds for a wide range of initial conditions, and handles missing data robustly. It is compared qualitatively and quantitatively to other algorithms using synthetic examples and real world data.",
"In this work, we analyze three different registration algorithms: Chamfer distance matching, the well-known iterated closest points (ICP) and an optic flow based registration. Their pairwise combination is investigated in the context of silhouette based pose estimation. It turns out that Chamfer matching and ICP used in combination do not only perform fairly well with small offset, but also deal with large offset significantly better than the other combinations. We show that by applying different optimized search strategies, the computational cost can be reduced by a factor eight. We further demonstrate the robustness of our method against simultaneous translation and rotation.",
"The authors describe a general-purpose, representation-independent method for the accurate and computationally efficient registration of 3-D shapes including free-form curves and surfaces. The method handles the full six degrees of freedom and is based on the iterative closest point (ICP) algorithm, which requires only a procedure to find the closest point on a geometric entity to a given point. The ICP algorithm always converges monotonically to the nearest local minimum of a mean-square distance metric, and the rate of convergence is rapid during the first few iterations. Therefore, given an adequate set of initial rotations and translations for a particular class of objects with a certain level of 'shape complexity', one can globally minimize the mean-square distance metric over all six degrees of freedom by testing each initial registration. One important application of this method is to register sensed data from unfixtured rigid objects with an ideal geometric model, prior to shape inspection. Experimental results show the capabilities of the registration algorithm on point sets, curves, and surfaces. >"
]
} |
1907.09133 | 2963230693 | Sensors producing 3D point clouds such as 3D laser scanners and RGB-D cameras are widely used in robotics, be it for autonomous driving or manipulation. Aligning point clouds produced by these sensors is a vital component in such applications to perform tasks such as model registration, pose estimation, and SLAM. Iterative closest point (ICP) is the most widely used method for this task, due to its simplicity and efficiency. In this paper we propose a novel method which solves the optimisation problem posed by ICP using stochastic gradient descent (SGD). Using SGD allows us to improve the convergence speed of ICP without sacrificing solution quality. Experiments using Kinect as well as Velodyne data show that, our proposed method is faster than existing methods, while obtaining solutions comparable to standard ICP. An additional benefit is robustness to parameters when processing data from different sensors. | Another avenue of research are methods that improve the ability of ICP to handle an unknown degree of overlap between the point clouds. This includes the use of a trimmed square cost function to estimate optimal transformation @cite_21 , rejection of corresponding point pairs based on a threshold defined by the standard deviation of point pair distances @cite_15 , or the use of a semi differential invariant for correspondence estimation @cite_9 . | {
"cite_N": [
"@cite_9",
"@cite_15",
"@cite_21"
],
"mid": [
"2561423101",
"2140711847",
"2144491133"
],
"abstract": [
"Despite the fact that original Iterative Closest Point(ICP) algorithm has been widely used for registration, itcannot tackle the problem when two point clouds are par-tially overlapping. Accordingly, this paper proposes a ro-bust approach for the registration of partially overlappingpoint clouds. Given two initially posed clouds, it firstlybuilds up bilateral correspondence and computes bidirec-tional distances for each point in the data shape. Based onthe ratio of bidirectional distances, the exponential functionis selected and utilized to calculate the probability value,which can indicate whether the point pair belongs to theoverlapping part or not. Subsequently, the probability val-ue can be embedded into the least square function for reg-istration of partially overlapping point clouds and a novelvariant of ICP algorithm is presented to obtain the optimalrigid transformation. The proposed approach can achievegood registration of point clouds, even when their overlappercentage is low. Experimental results tested on public da-ta sets illustrate its superiority over previous approaches onrobustness.",
"The problem of geometric alignment of two roughly preregistered, partially overlapping, rigid, noisy 3D point sets is considered. A new natural and simple, robustified extension of the popular Iterative Closest Point (ICP) algorithm (Besl and McKay, 1992) is presented, called the Trimmed ICP (TrICP). The new algorithm is based on the consistent use of the least trimmed squares (LTS) approach in all phases of the operation. Convergence is proved and an efficient implementation is discussed. TrICP is fast, applicable to overlaps under 50 , robust to erroneous measurements and shape defects, and has easy-to-set parameters. ICP is a special case of TrICP when the overlap parameter is 100 . Results of testing the new algorithm are shown.",
"In this paper we present a novel on-line method to recursively align point clouds. By considering each point together with the local features of the surface (normal and curvature), our method takes advantage of the 3D structure around the points for the determination of the data association between two clouds. The algorithm relies on a least squares formulation of the alignment problem, that minimizes an error metric depending on these surface characteristics. We named the approach Normal Iterative Closest Point (NICP in short). Extensive experiments on publicly available benchmark data show that NICP outperforms other state-of-the-art approaches."
]
} |
1907.09189 | 2963689651 | This paper is concerned with evaluating different multiagent learning (MAL) algorithms in problems where individual agents may be heterogenous, in the sense of utilizing different learning strategies, without the opportunity for prior agreements or information regarding coordination. Such a situation arises in ad hoc team problems, a model of many practical multiagent systems applications. Prior work in multiagent learning has often been focussed on homogeneous groups of agents, meaning that all agents were identical and a priori aware of this fact. Also, those algorithms that are specifically designed for ad hoc team problems are typically evaluated in teams of agents with fixed behaviours, as opposed to agents which are adapting their behaviours. In this work, we empirically evaluate five MAL algorithms, representing major approaches to multiagent learning but originally developed with the homogeneous setting in mind, to understand their behaviour in a set of ad hoc team problems. All teams consist of agents which are continuously adapting their behaviours. The algorithms are evaluated with respect to a comprehensive characterisation of repeated matrix games, using performance criteria that include considerations such as attainment of equilibrium, social welfare and fairness. Our main conclusion is that there is no clear winner. However, the comparative evaluation also highlights the relative strengths of different algorithms with respect to the type of performance criteria, e.g., social welfare vs. attainment of equilibrium. | Harsanyi pioneered the study of incomplete information games. In his 1967 paper @cite_31 , he describes the , a game in which players have beliefs about missing information. He develops the concept of the Bayesian Nash equilibrium @cite_34 in which each player plays a best response against the other players, based on the personal beliefs of the player. Jordan @cite_17 showed that, for any repeated game, if the players play a Bayesian Nash equilibrium in each repetition, and if the personal beliefs of the players satisfy certain conditions, then this will converge to a true Nash equilibrium. | {
"cite_N": [
"@cite_31",
"@cite_34",
"@cite_17"
],
"mid": [
"2102794165",
"2097540415",
"2139774323"
],
"abstract": [
"Part I of this paper has described a new theory for the analysis of games with incomplete information. It has been shown that, if the various players' subjective probability distributions satisfy a certain mutual-consistency requirement, then any given game with incomplete information will be equivalent to a certain game with complete information, called the “Bayes-equivalent” of the original game, or briefly a “Bayesian game.” Part II of the paper will now show that any Nash equilibrium point of this Bayesian game yields a “Bayesian equilibrium point” for the original game and conversely. This result will then be illustrated by numerical examples, representing two-person zero-sum games with incomplete information. We shall also show how our theory enables us to analyze the problem of exploiting the opponent's erroneous beliefs. However, apart from its indubitable usefulness in locating Bayesian equilibrium points, we shall show it on a numerical example the Bayes-equivalent of a two-person cooperative game that the normal form of a Bayesian game is in many cases a highly unsatisfactory representation of the game situation and has to be replaced by other representations e.g., by the semi-normal form. We shall argue that this rather unexpected result is due to the fact that Bayesian games must be interpreted as games with “delayed commitment” whereas the normal-form representation always envisages a game with “immediate commitment.”",
"Two players are about to play a discounted infinitely repeated bimatrix game. Each player knows his own payoff matrix and chooses a strategy which is a best response to some private beliefs over strategies chosen by his opponent. If both players' beliefs contain a grain of truth (each assigns some positive probability to the strategy chosen by the opponent), then they will eventually (a) accurately predict the future play of the game and (b) play a Nash equilibrium of the repeated game. An immediate corollary is that in playing a Harsanyi-Nash equilibrium of a discounted repeated game of incomplete information about opponents' payoffs, the players will eventually play an equilibrium of the real game as if they had complete information.",
"(This article originally appeared in Management Science, November 1967, Volume 14, Number 3, pp. 159-182, published by The Institute of Management Sciences.) The paper develops a new theory for the analysis of games with incomplete information where the players are uncertain about some important parameters of the game situation, such as the payoff functions, the strategies available to various players, the information other players have about the game, etc. However, each player has a subjective probability distribution over the alternative possibilities. In most of the paper it is assumed that these probability distributions entertained by the different players are mutually \"consistent,\" in the sense that they can be regarded as conditional probability distributions derived from a certain \"basic probability distribution\" over the parameters unknown to the various players. But later the theory is extended also to cases where the different players' subjective probability distributions fail to satisfy this consistency assumption. In cases where the consistency assumption holds, the original game can be replaced by a game where nature first conducts a lottery in accordance with the basic probability distribution, and the outcome of this lottery will decide which particular subgame will be played, i.e., what the actual values of the relevant parameters will be in the game. Yet, each player will receive only partial information about the outcome of the lottery, and about the values of these parameters. However, every player will know the \"basic probability distribution\" governing the lottery. Thus, technically, the resulting game will be a game with complete information. It is called the Bayes-equivalent of the original game. Part I of the paper describes the basic model and discusses various intuitive interpretations for the latter. Part II shows that the Nash equilibrium points of the Bayes-equivalent game yield \"Bayesian equilibrium points\" for the original game. Finally, Part III considers the main properties of the \"basic probability distribution.\""
]
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.