Author: Yuchao Gu
E-mail: 2015014178@buct.edu.cn
Date: 2018-05-27
Description: The code is an pytorch implementation of 《Retinal Vessel Segmentation in Fundoscopic Images with Generative Adversarial Networks》
DRIVE: Digital Retinal Images for Vessel Extraction you can download the train and test data from this server. You can also find data in the eyedata folder.
The dataset contains 20 training images, the first step of my pre-processing is randomly cropping into 512*512. The second step is to randomly change brightness ,contrast and hue of the train image. I implement this method in my code, so you can be convenient to use it. Further more, a gan-based method of generating retina images can be used as an extra data source.
python train.py
This code depends on the following libraries:
- Python 3.6
- Pytorch
- PIL
vessel gan
│
├── eyedata # drive data
│
├── gycutils # my utils for data augmentation
│
├── Criterion.py # generate and store precison,recall curve
│
├── datasets.py # dataset for dataloader
│
├── gan.py # generative adversial network for vessel segmentation
│
├── train.py # train code
│
├── transform.py
│
└── readme.md # introduce to this project