Cross-domain disentanglement network
Switch branches/tags
Nothing to show
Clone or download
Latest commit 2d189e1 Nov 4, 2018
Type Name Latest commit message Commit time
Failed to load latest commit information.
DATA/MNISTCDCB Add example images Nov 4, 2018
tools added WGAN-GP loss Feb 2, 2018
.gitignore ignore folders Feb 13, 2018
LICENSE.txt Code cleanup Nov 4, 2018 Update Nov 4, 2018 Code cleanup Nov 4, 2018 Code cleanup Nov 4, 2018 Code cleanup Nov 4, 2018 Code cleanup Nov 4, 2018 Code cleanup Nov 4, 2018 main file Nov 2, 2018 init Jan 30, 2018 Code cleanup Nov 4, 2018

Cross-domain disentantanglement network

Code for the paper "Image-to-image translation for cross-domain disentanglement", NIPS 2018.

Cross-domain disentanglement network

Based on this pix2pix implementation by Christopher Hesse, extensively explained in this article.


Please follow the setup described here. Tested with Tensorflow 1.8.0.

See DATA/MNISTCDCB/ for example images of our MNIST-CD/CB dataset.


In order to train a MODEL using DATA, run

python \
  --mode train \ 
  --output_dir checkpoints/MODEL \ 
  --input_dir DATA/train/  

Once the model finished training, it can be tested by running

python \ 
  --mode test \ 
  --output_dir test/MODEL \
  --checkpoint checkpoints/MODEL \
  --input_dir DATA/test/  

In order to extract disentangled features for other tasks (e.g. cross-domain retrieval), run

python \ 
  --mode features \ 
  --output_dir features/MODEL \ 
  --checkpoint checkpoints/MODEL \ 
  --input_dir DATA/test/  


Please, cite the following paper if you use this code:

  title={Image-to-image translation for cross-domain disentanglement},
  author={Gonzalez-Garcia, Abel and van de Weijer, Joost and Bengio, Yoshua},