This repository contains the code associated to the paper "Multi-style conversion for semantic segmentation of lesions in fundus images by adversarial attacks"
pip install .To train the model, you will have to adjust the config file to correspond the paths to the five datasets:
1. IDRID
2. MESSIDOR
3. FGADR
4. DDR
5. RETINAL LESIONS
We focus on the semantic segmentations of:
1. Cotton Wool Spot
2. Exudates
3. Hemmorrhages
4. Microaneurysms
For an easier use of the models, we refer to cleaned-up version of the code provided in the fundus lesions toolkit.
Modifying an image by adversarial attacks to move from RETINAL LESIONS to IDRiD style.
Visualization of the values added by gradient attack.
