Skip to content

A concise PyTorch implementation of CUT (Contrastive unpaired image-to-image translation)

Notifications You must be signed in to change notification settings

Kaminyou/Contrastive-Unpaired-Translation-CUT-

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

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.

About

A concise PyTorch implementation of CUT (Contrastive unpaired image-to-image translation)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages