Awesome

Self-Supervised Vessel Segmentation via Adversarial Learning

If you find our dataset, code or paper useful, please cite as

@inproceedings{SSVS,
    title = {Self-Supervised Vessel Segmentation via Adversarial Learning},
    author = {Yuxin Ma, Yang Hua, Hanming Deng, Tao Song, Hao Wang, Zhengui Xue,Heng Cao, Ruhui Ma and Haibing Guan},
    booktitle = {IEEE International Conference on Computer Vision (ICCV)},
    year = {2021}
}

XCAD dataset

You can download the XCAD dataset from Dropbox or BaiduNetdisk with the key neia.

Code

Environment

Follow ssv_env.yml to create the environment.

Data Preparation

Put the downloaded dataset to dataset/ssv.

XCAD/train/* ->datasets/ssv/*
XCAD/test/images -> datasets/ssv/testB
XCAD/test/masks -> datasets/ssv/testA

Put some background images (e.g.,XCAD/train/trainC) to datasets/ssv/testC. Using fractal.py to synthesis fractals and put it in datasets/ssv/trainA.

Train

Follow the scripts of arun_train.sh to train the model.

Test and Evaluation

Use arun_test.sh for inference, and use eval/test.sh to get evaluation results.