Awesome
CPC-Trans
Code for the MICCAI 2022 (early accepted) paper: "Toward Clinically Assisted Colorectal Polyp Recognition via Structured Cross-modal Representation Consistency"
<p align="center"> <img align="middle" src="./assets/overview.png" style="width:80%" alt="The main figure"/> </p>
Dataset
We only provide the public part [link] with its bbox we labeled [link] of the CPC-paired Dataset while the another part is private. Thanks for understanding. And you should label the bounding box of the polyp and crop for preprocessing.
Installation
pip install -r requirements.txt
Download Pretrained Weights
We provide the pretrained weights of the model which were mentioned in the paper.
- Base Weights (vit_small_patch16_224_in21k) link
For other model sizes, please refer to this table and download on the Internet or contact us to provide.
<p align="center"> <img align="middle" src="./assets/variants.png" style="width:50%" alt="The main figure"/> </p>- Smaller Weights (vit_tiny_patch16_224_in21k)
- Larger Weights (vit_base_patch16_224_in21k)
Run
we set the argument "fold" (default: 0) for k-fold cross validation, you can omit it if unnecessary.
python main.py --data_path /the/data/path/ --weights /the/pretrained/weights/path/
Citation
If you use any part of this code and pretrained weights for your own purpose, please cite our paper.
@InProceedings{Ma-CPC-Trans,
author={Ma, Weijie and Zhu, Ye and Zhang, Ruimao and Yang, Jie and Hu, Yiwen and Li, Zhen and Xiang, Li},
title={Toward Clinically Assisted Colorectal Polyp Recognition via Structured Cross-Modal Representation Consistency},
booktitle={Medical Image Computing and Computer Assisted Intervention -- MICCAI 2022},
year={2022},
publisher={Springer Nature Switzerland},
pages={141--150}
}