Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<div align="left">
|
2 |
+
<h1 style="font-size: 2em;">π RoomSpace Benchmark</h1>
|
3 |
+
π <a href="https://roomspace-benchmark.web.app/">Homepage</a> β’
|
4 |
+
π <a href="https://github.com/Fangjun-Li/RoomSpace">GitHub</a> β’
|
5 |
+
π <a href="https://archive.researchdata.leeds.ac.uk/1293/">Dataset</a> β’
|
6 |
+
π <a href="https://arxiv.org/abs/2405.15064">Paper</a>
|
7 |
+
πΉοΈ <a href="https://colab.research.google.com/drive/1fAK8J1UHAjMm-mNVsuzIbEZd-SZG6bX-?usp=sharing">Colab</a>
|
8 |
+
</div>
|
9 |
+
<br/>
|
10 |
+
|
11 |
+
|
12 |
+
## Branch Structure
|
13 |
+
|
14 |
+
This repository contains various branches, each representing different versions and sizes of the RoomSpace dataset. For example, the `SD-100_v1.0` branch includes 100 test examples under version 1.0.
|
15 |
+
|
16 |
+
```
|
17 |
+
Data/
|
18 |
+
βββ main/
|
19 |
+
βββ SD-100_v1.0/
|
20 |
+
βββ SD-1K_v1.0/
|
21 |
+
βββ SD-10K_v1.0/
|
22 |
+
```
|
23 |
+
|
24 |
+
- `main`: for the Dataset Viewer, contains the `n5_m4_d144` directory for SD-1K_v1.0.
|
25 |
+
- `SD-100_v1.0`: Contains data for 100 room samples.
|
26 |
+
- `SD-1K_v1.0`: Contains data for 1,000 room samples, with the first 100 rooms identical to those in `SD-100_v1.0`.
|
27 |
+
- `SD-10K_v1.0`: Contains data for 10,000 room samples, with the first 1,000 rooms identical to those in `SD-1K_v1.0`.
|
28 |
+
|
29 |
+
## Directory Structure
|
30 |
+
|
31 |
+
The repository is organized as follows:
|
32 |
+
|
33 |
+
```
|
34 |
+
Revision/
|
35 |
+
βββ n3_m2_d144/
|
36 |
+
βββ n4_m3_d144/
|
37 |
+
```
|
38 |
+
|
39 |
+
Each `n*_m*_d*` subdirectory includes a test parquet file containing stories, questions, and answers derived from the images and a Constraint Satisfaction Problem (CSP) reasoner. The parameters `n`, `m`, and `d` represent the object number, constraint density, and domain size, respectively.
|
40 |
+
|
41 |
+
### Images
|
42 |
+
|
43 |
+
Each example includes two images showing different views of the rooms:
|
44 |
+
|
45 |
+
- `image_top_down`: Top-down images of the rooms.
|
46 |
+
- `image_north_facing`: North-facing rectangular images, assuming a robot stands at the south door, facing north.
|
47 |
+
|
48 |
+
### Questions & Answers
|
49 |
+
|
50 |
+
There are a total of 26 QA pairs, categorized as follows:
|
51 |
+
|
52 |
+
- **Visual QA**: (image_top_down + question_image) -- answer_image_td, (image_north_facing + question_image) -- answer_image_sd
|
53 |
+
- **Layout Setting**: (question_td_layout_yn -- answer_td_layout_yn), (question_td_layout_fr -- answer_td_layout_fr)
|
54 |
+
- **TPP Setting**: (question_td_tpp_yn -- answer_td_tpp_yn), (question_td_tpp_fr -- answer_td_tpp_fr)
|
55 |
+
- **O2 Setting**: (question_td_o2_yn -- answer_td_o2_yn), (question_td_o2_fr -- answer_td_o2_fr), (question_sd_o2_yn -- answer_sd_o2_yn), (question_sd_o2_fr -- answer_sd_o2_fr)
|
56 |
+
- **O2+D2 Setting**: (question_td_o2_d2_yn -- answer_td_o2_d2_yn), (question_td_o2_d2_fr -- answer_td_o2_d2_fr), (question_sd_o2_d2_yn -- answer_sd_o2_d2_yn), (question_sd_o2_d2_fr -- answer_sd_o2_d2_fr)
|
57 |
+
- **O2+D3 Setting**: (question_td_o2_d3_yn -- answer_td_o2_d3_yn), (question_td_o2_d3_fr -- answer_td_o2_d3_fr), (question_sd_o2_d3_yn -- answer_sd_o2_d3_yn), (question_sd_o2_d3_fr -- answer_sd_o2_d3_fr)
|
58 |
+
- **Layout + O2 + D2 Setting**: (question_td_o2_d2_layout_yn -- answer_td_o2_d2_layout_yn), (question_td_o2_d2_layout_fr -- answer_td_o2_d2_layout_fr), (question_sd_o2_d2_layout_yn -- answer_sd_o2_d2_layout_yn), (question_sd_o2_d2_layout_fr -- answer_sd_o2_d2_layout_fr)
|
59 |
+
- **Layout + O2 + D3 Setting**: (question_td_o2_d3_layout_yn -- answer_td_o2_d3_layout_yn), (question_td_o2_d3_layout_fr -- answer_td_o2_d3_layout_fr), (question_sd_o2_d3_layout_yn -- answer_sd_o2_d3_layout_yn), (question_sd_o2_d3_layout_fr -- answer_sd_o2_d3_layout_fr)
|
60 |
+
|
61 |
+
|
62 |
+
|
63 |
+
### Question Types
|
64 |
+
|
65 |
+
- **'fr'**: Find Relation
|
66 |
+
- **'yn'**: Yes/No
|
67 |
+
|
68 |
+
|
69 |
+
### Points of View
|
70 |
+
|
71 |
+
- **'td'**: Top-Down View
|
72 |
+
- **'sd'**: North Facing View (from the south door)
|
73 |
+
|
74 |
+
|
75 |
+
---
|
76 |
+
## Relevant Paper
|
77 |
+
|
78 |
+
[Fangjun Li, David Hogg, and Anthony Cohn. "Reframing Spatial Reasoning Evaluation in Language Models: A Real-World Simulation Benchmark for Qualitative Reasoning." IJCAI, 2024](https://arxiv.org/abs/2405.15064).
|