Awesome
Weakly Supervised Polyp Frame Detection
[MICCAI'22 Early Accept] Contrastive Transformer-based Multiple Instance Learning for Weakly Supervised Polyp Frame Detection
by Yu Tian, Guansong Pang, Fengbei Liu, Yuyuan Liu, Chong Wang, Yuanhong Chen, Johan W Verjans, Gustavo Carneiro.
Dataset
Please download the pre-processed i3d features of the dataset through this link
The original colonoscopy videos can be found in this link.
Training
After downloading the dataset and extracting the I3D features using this repo, simply run the following command:
python main_transformer.py
Inference
For inference, after setting the path of the best checkpoint, then run the following command:
python inference.py
Citation
If you find this repo useful for your research, please consider citing our paper:
@inproceedings{tian2022contrastive,
title={Contrastive Transformer-Based Multiple Instance Learning for Weakly Supervised Polyp Frame Detection},
author={Tian, Yu and Pang, Guansong and Liu, Fengbei and Liu, Yuyuan and Wang, Chong and Chen, Yuanhong and Verjans, Johan and Carneiro, Gustavo},
booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
pages={88--98},
year={2022},
organization={Springer}
}
If you use the dataset, please also consider citing the papers below:
@inproceedings{ma2021ldpolypvideo,
title={Ldpolypvideo benchmark: A large-scale colonoscopy video dataset of diverse polyps},
author={Ma, Yiting and Chen, Xuejin and Cheng, Kai and Li, Yang and Sun, Bin},
booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
pages={387--396},
year={2021},
organization={Springer}
}
@article{borgli2020hyperkvasir,
title={HyperKvasir, a comprehensive multi-class image and video dataset for gastrointestinal endoscopy},
author={Borgli, Hanna and Thambawita, Vajira and Smedsrud, Pia H and Hicks, Steven and Jha, Debesh and Eskeland, Sigrun L and Randel, Kristin Ranheim and Pogorelov, Konstantin and Lux, Mathias and Nguyen, Duc Tien Dang and others},
journal={Scientific data},
volume={7},
number={1},
pages={1--14},
year={2020},
publisher={Nature Publishing Group}
}