search

Retinal-Research / OTE-GAN

Code for the paper "OPTIMAL TRANSPORT GUIDED UNSUPERVISED LEARNING FOR ENHANCING LOW-QUALITY RETINAL IMAGES"
MIT License
72 stars 0 forks source link

How can I use the Model #2

Closed Body123 closed 10 months ago

Body123 commented 10 months ago

is there a link to the pre-trained weights and how I can use the repository

ChongQingNoSubway commented 10 months ago

All our model training code was in /OTE-GAN, Without pre-trained weights There are two experiments on unsupervised and degradation dataset settings.

Unsupervised setting prepared the dataset from Quality label [https://github.com/HzFu/EyeQ] EYEPACS [https://www.kaggle.com/c/diabetic-retinopathy-detection/data] -> Change the root dictionary in train.py -> parser.add_argument("--root", default="dataset/preprocessed/ALLDATA", type=str) parser.add_argument("--file_dir",default="dataset/Label_EyeQ_train.csv", type=str) -> Run /OTE-GAN/train.py (Other baseline models followed the same steps)

Degradation setting processing from https://github.com/joanshen0508/Fundus-correction-cofe-Net -> Change the root dictionary in the train_full.py parser.add_argument("--root", default="dataset/degratation/pre", type=str) parser.add_argument("--file_dir",default="dataset/train.csv", type=str) -> Run /OTE-GAN/train_full.py (Other baseline models followed the same steps)