Self-Supervised Correction Learning for Semi-Supervised Biomedical Image Segmentation

Introduction

This repository contains the PyTorch implementation of:

Self-Supervised Correction Learning for Semi-Supervised Biomedical Image Segmentation, MICCAI 2021.

Requirements

• torch
• torchvision
• tqdm
• opencv
• scipy
• skimage
• PIL
• numpy

Usage

1. Training

python main.py --mode train --manner semi --ratio 2 --root {project_path} --dataset polyp --polyp {data_path}

2. Inference

python main.py --mode test --manner test --load_ckpt checkpoint --root {project_path} --dataset polyp --polyp {data_path}

Citation

If you feel this work is helpful, please cite our paper

@inproceedings{zhang2021self,
title={Self-supervised Correction Learning for Semi-supervised Biomedical Image Segmentation},
author={Zhang, Ruifei and Liu, Sishuo and Yu, Yizhou and Li, Guanbin},
booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
pages={134--144},
year={2021},
organization={Springer}
}