Skip to content
Branch: master
Find file History
Thibault de Boissiere
Thibault de Boissiere Remove deprecated mode kwarg
Latest commit 1e2a864 Nov 19, 2017
Permalink
Type Name Latest commit message Commit time
..
Failed to load latest commit information.
data Add WassersteinGAN Feb 1, 2017
figures Add more hyperparam in argparse and cleanup Feb 2, 2017
models
src
README.md Update to Keras 2 Oct 30, 2017

README.md

Generative Adversarial Networks

Keras implementation of WassersteinGAN.

Sources:

Requirements

python modules

numpy==1.13.3 natsort==5.1.0 matplotlib==2.0.2 opencv_python==3.3.0.10 scipy==1.0.0 tqdm==4.17.0 Keras==2.0.8 parmap==1.5.1 h5py==2.7.0 Theano==0.9.0 or tensorflow==1.3.0

Part 1. Processing the data

Follow these instructions.

Part 2. Running the code

Follow these instructions

Part 3. Example results

CelebA example results

figure figure

For each image:

  • The first 2 rows are generated images
  • The last 2 rows are real images

MoG

Results on the unrolled GAN paper to dataset:

figure

You can’t perform that action at this time.