This repository is implementation of ICCV2019 paper "InstaBoost: Boosting Instance Segmentation Via Probability Map Guided Copy-Pasting". Our paper has been released on arXiv https://arxiv.org/abs/1908.07801.
To install InstaBoost, use this command.
pip install instaboost
The detail implementation can be found here
.
Because InstaBoost depends on matting package here, we highly recommend users to use python3.5 or 3.6, OpenCV 2.4 to avoid some errors. Envrionment setting instructions can be found here.
Video demo for InstaBoost: https://www.youtube.com/watch?v=iFsmmHUGy0g
Currently we have integrated InstaBoost into three open implementations: mmdetection, detectron and yolact.
-
mmdetection: Checkout mmdetection.
-
detectron: Checkout detectron.
-
yolact: Checkout yolact
Since these frameworks may continue updating, codes in this repo may be a little different from their current repo.
It is easy to integrate InstaBoost into your framework. You can refer to instructions of our implementations on mmdetection, detectron and yolact
To change InstaBoost Configurations, users can use function InstaBoostConfig
.
Results and models are available in the Model zoo. More models are coming!
If you use this toolbox or benchmark in your research, please cite this project.
@article{Fang2019InstaBoost,
author = {Fang, Hao-Shu and Sun, Jianhua and Wang, Runzhong and Gou, Minghao and Li, Yong-Lu and Lu, Cewu},
title = {InstaBoost: Boosting Instance Segmentation Via Probability Map Guided Copy-Pasting},
journal={arXiv preprint arXiv:1908.07801},
year = {2019}
}
Please also cite mmdetection, detectron and yolact if you use the corresponding codes.
Our detection and instance segmentation framework is based on mmdetecion, detectron and yolact.