Skip to content

DrZhanYing/conditional-dcgan-keras

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

23 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

KERAS CONDITIONAL GAN

Implementation of conditional DCGAN https://arxiv.org/abs/1411.1784 with keras.

Implements the improvements and architecture of https://arxiv.org/pdf/1611.07004v1.pdf.

Code borrows from the Keras DCGAN https://github.com/jacobgil/keras-dcgan and the tensorflow conditional GAN https://github.com/ppwwyyxx/tensorpack/blob/master/examples/GAN/Image2Image.py.


This assumes theano ordering.


Usage

Training:

KERAS_BACKEND=theano THEANO_FLAGS=optimizer=fast_compile,device=cuda0,floatX=float32 ./conditional_gan.py --data train

KERAS_BACKEND=theano THEANO_FLAGS=optimizer=fast_compile,device=cuda0,floatX=float32 ./conditional_gan.py --data train

Image generation:

KERAS_BACKEND=theano THEANO_FLAGS=optimizer=fast_compile,device=cuda0,floatX=float32 ./conditional_gan.py --data test

KERAS_BACKEND=theano THEANO_FLAGS=optimizer=fast_compile,device=cuda0,floatX=float32 ./conditional_gan.py --data test


Result

generated images (conditioned on RGB) :

generated_image.png

train process :

training_process1.gif training_process2.gif training_process3.gif


About

Keras implementation of the conditional GAN.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%