EDGSCOUT / Entity

EntitySeg Toolbox: Towards precise and open-world image segmentation

EntitySeg Toolbox: Towards precise and open-world image segmentation

EntitySeg is an open source toolbox which towards precise and open-world image segmentation on top of detectron2. All works related to image segmentation from our group are open-sourced here.

To date, EntitySeg implements the following algorthms:

• Open-World Entity Segmentation
• High Quality Segmentation for Ultra High-resolution Images ---code to be released
• CaSP: Class-agnostic Semi-Supervised Pretraining for Detection and Segmentation ---code to be released
• Scale-aware Automatic Augmentation for Object Detection ---code to be merged

Usage

Please refer to the README.md of each project. All projects shared the similar detectron2 base code and could support each other.

Citing Ours

@article{qi2021open,
  title={Open-World Entity Segmentation},
  author={Qi, Lu and Kuen, Jason and Wang, Yi and Gu, Jiuxiang and Zhao, Hengshuang and Lin, Zhe and Torr, Philip and Jia, Jiaya},
  journal={arXiv preprint arXiv:2107.14228},
  year={2021}
}

@inproceedings{chen2021scale,
  title={Scale-aware Automatic Augmentation for Object Detection},
  author={Chen, Yukang and Li, Yanwei and Kong, Tao and Qi, Lu and Chu, Ruihang and Li, Lei and Jia, Jiaya},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={9563--9572},
  year={2021}
}