Contrastive Unpaired Translation (CUT) concise implementation

A simplified and concise implementation of Contrastive Learning for Unpaired Image-to-Image Translation (ECCV 2020)

Knowledgement

Please refer to the official implementation here. This code is a simplified version revised from wilbertcaine's implementation.

Usage (Training)

Suppose you would like to transfer image from source domain X to the target domain Y.

1. Please split your data X into training and testing set. Namely, split X into X_train and X_test.
2. Put X_train into ./data/trainX.
3. Put X_test into ./data/testX.
4. Put Y into ./data/trainY.
5. Modify ./config.yaml if you would like to adjust some setting, or just keep the default setting.
6. Execute python3 train.py. You can use python3 train.py --verbose to see more info during training.
7. Some transfered examples will be generated during training. Please check the ./experiments/$experiment_name/train/ folder.

Usage (Inference)

python3 inference.py

The transfered images will be stored in ./experiments/$experiment_name/test/ folder.

Environment

• Python 3.8.6
• All the required packages are listed in the requirements.txt.