---
title: README
emoji: ⚡
colorFrom: yellow
colorTo: green
sdk: static
pinned: false
---

<div align="center">
   <p><em>Easily load and transform datasets for object detection.</em></p>
</div>
<br></p>
<hr>
<p><strong>Documentation</strong>: <a href="https://blinjrm.github.io/detection-datasets/">https://blinjrm.github.io/detection-datasets/</a></p>
<p><strong>Source Code</strong>: <a href="https://github.com/blinjrm/detection-datasets">https://github.com/blinjrm/detection-datasets</a></p>
<p><strong>Datasets on Hugging Face Hub</strong>: <a href="https://huggingface.co/detection-datasets">https://huggingface.co/detection-datasets</a></p>
<hr>
<p><br></p>
<p><code>detection_datasets</code> aims to make it easier to work with detection datasets.  </p>
<p>This library works alongside the <a href="https://huggingface.co/detection-datasets">Detection dataset</a> organisation on the 🤗 Hub, where some detection datasets have been uploaded in the format expected by the library, and are ready to use.  </p>
<p>The main features are:</p>
<ul>
   <li>
      <strong>Read</strong> the dataset :
      <ul>
         <li>From disk if it has already been downloaded.</li>
         <li>Directly from the Hugging Face Hub if it <a href="https://huggingface.co/detection-datasets">already exist</a>.</li>
      </ul>
   </li>
   <li>
      <strong>Transform</strong> the dataset:
      <ul>
         <li>Select a subset of data.</li>
         <li>Remap categories.</li>
         <li>Create new train-val-test splits.</li>
      </ul>
   </li>
   <li><strong>Visualize</strong> the annotations and images.</li>
   <li>
      <strong>Write</strong> the dataset:
      <ul>
         <li>To disk, selecting the target detection format: <code>COCO</code>, <code>YOLO</code> and more to come.</li>
         <li>To the Hugging Face Hub for easy reuse in a different environment and share with the community.</li>
      </ul>
   </li>
</ul>
<p><br></p>
<div align="center">
<hr>
<p><em>Read the <a href="https://github.com/blinjrm/detection-datasets/blob/main/README.md#getting-started">quick start</a>, or directly jump to the tutorials:</em></p>
<table>
   <thead>
      <tr>
         <th>Goal</th>
         <th style="text-align:center">Tutorial</th>
         <th style="text-align:center">Colab</th>
      </tr>
   </thead>
   <tbody>
      <tr>
         <td>Load from disk and upload to the Hub</td>
         <td style="text-align:center"><a href="https://blinjrm.github.io/detection-datasets/tutorials/1_Read/">Open in the docs</a></td>
         <td style="text-align:center"><a href="https://colab.research.google.com/github/blinjrm/detection-datasets/blob/main/docs/tutorials/1_Read.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"></a></td>
      </tr>
      <tr>
         <td>Load from the Hub and transform</td>
         <td style="text-align:center"><a href="https://blinjrm.github.io/detection-datasets/tutorials/2_Transform/">Open in the docs</a></td>
         <td style="text-align:center"><a href="https://colab.research.google.com/github/blinjrm/detection-datasets/blob/main/docs/tutorials/2_Transform.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"></a></td>
      </tr>
   </tbody>
</table>
<hr>