Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/Shengcao-Cao/HASSOD
[NeurIPS 2023] HASSOD: Hierarchical Adaptive Self-Supervised Object Detection
https://github.com/Shengcao-Cao/HASSOD
Last synced: 3 months ago
JSON representation
[NeurIPS 2023] HASSOD: Hierarchical Adaptive Self-Supervised Object Detection
- Host: GitHub
- URL: https://github.com/Shengcao-Cao/HASSOD
- Owner: Shengcao-Cao
- License: apache-2.0
- Created: 2023-10-28T04:00:49.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2024-02-08T01:23:37.000Z (9 months ago)
- Last Synced: 2024-06-19T02:49:23.671Z (5 months ago)
- Language: Python
- Homepage: https://hassod-neurips23.github.io/
- Size: 7.76 MB
- Stars: 44
- Watchers: 2
- Forks: 3
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- Awesome-Transformer-Attention - [Paper - Cao/HASSOD)][[Website](https://hassod-neurips23.github.io/)] (Detection / Other Detection Tasks)
README
# HASSOD: Hierarchical Adaptive Self-Supervised Object Detection
This is the official PyTorch implementation of our *NeurIPS 2023* paper:
**HASSOD: Hierarchical Adaptive Self-Supervised Object Detection**
[[Project Page]](https://hassod-neurips23.github.io/) [[Paper-arXiv]](https://arxiv.org/abs/2402.03311) [[Paper-OpenReview]](https://openreview.net/pdf?id=sqkGJjIRfG) [[Video-YouTube]](https://www.youtube.com/watch?v=s8u7tEKg5ew) [[Video-Bilibili]](https://www.bilibili.com/video/BV1pg4y1Z7CK)
[Shengcao Cao](https://shengcao-cao.github.io/), [Dhiraj Joshi](https://research.ibm.com/people/dhiraj-joshi), [Liang-Yan Gui](https://cs.illinois.edu/about/people/faculty/lgui), [Yu-Xiong Wang](https://yxw.web.illinois.edu/)
## 🔎 Overview
![HASSOD-gif](assets/HASSOD.gif)
HASSOD is a fully self-supervised approach for object detection and instance segmentation, demonstrating a significant improvement over the previous state-of-the-art methods by discovering a more comprehensive range of objects. Moreover, HASSOD understands the part-to-whole object composition like humans do, while previous methods cannot. Notably, we improve class-agnostic Mask AR from 20.2 to 22.5 on LVIS, and from 17.0 to 26.0 on SA-1B.
## 🛠️ Instructions
To use our code and reproduce the results, please follow these detailed documents step by step:
- [Preparation](https://github.com/Shengcao-Cao/HASSOD/blob/main/preparation.md): Prepare the environment, data, and pre-trained models
- [Reproduction](https://github.com/Shengcao-Cao/HASSOD/blob/main/reproduction.md): Produce pseudo-labels and train the object detector (download links included for our pseudo-labels and model)
- [Demo](https://github.com/Shengcao-Cao/HASSOD/blob/main/demo.md): Once the preparation is finished, you can try out the demo code and test our model on any image.## 🙏 Acknowledgements
Our code is developed based on the following repositories:
- [CutLER](https://github.com/facebookresearch/CutLER)
- [Unbiased Teacher](https://github.com/facebookresearch/unbiased-teacher)
- [Detectron2](https://github.com/facebookresearch/detectron2)We greatly appreciate their open-source work!
## ⚖️ License
This project is released under the Apache 2.0 license. Other codes from open source repository follows the original distributive licenses.## 🌟 Citation
If you find our research interesting or use our code, data, or model in your research, please consider citing our work.
```
@inproceedings{cao2023hassod,
title={{HASSOD}: Hierarchical Adaptive Self-Supervised Object Detection},
author={Cao, Shengcao and Joshi, Dhiraj and Gui, Liangyan and Wang, Yu-Xiong},
booktitle={NeurIPS},
year={2023}
}
```