Unsupervised Domain Adaptation for Cross-Modality Retinal Vessel Segmentation via Disentangling Representation Style Transfer and Collaborative Consistency Learning

Pytorch implementation for our unsupervised domain adaptation framework with application to retinal vessel segmentation. We use style transfer and collaborative consistency learning to train a segmentation model on the target domain.

Paper

Please cite our paper if you find the code useful for your research.

@article{peng2022unsupervised,
  title={Unsupervised Domain Adaptation for Cross-Modality Retinal Vessel Segmentation via Disentangling Representation Style Transfer and Collaborative Consistency Learning},
  author={Peng, Linkai and Lin, Li and Cheng, Pujin and Huang, Ziqi and Tang, Xiaoying},
  journal={arXiv preprint},
  url={arXiv:2201.04812},
  year={2022}
}

Example Results

Usage

Prerequisite

• Python 3.7+
• Pytorch 1.4.0

Acknowledgement

Code adapted from DRIT.