Tensorflow Generative Adversarial Network unsupervised-learning Keras Deep learning Python Machine learning
Need help with GAN-Sandbox?
Click the “chat” button below for chat support from the developer who created it, or find similar developers for support.
mjdietzx

Description

Vanilla GAN implemented on top of keras/tensorflow enabling rapid experimentation & research. Branches correspond to implementations of stable GAN variations (i.e. ACGan, InfoGAN) and other promising variations of GANs like conditional and Wasserstein.

204 Stars 73 Forks MIT License 20 Commits 1 Opened issues

Services available

Need anything else?

GAN-Sandbox

Standard GAN implemented on top of keras/tensorflow enabling rapid experimentation & research. Branches correspond to stable implementations of GAN architectures (i.e. ACGan, InfoGAN, Improved wGAN) and other promising variations of GANs (i.e. GAN hacks, local adversarial loss, etc...).

Guidelines

The

master
branch serves as a simple, clean and robust starting point for GAN R&D. Contributions are encouraged in the form of new branches and/or improvements to
master
. Ideally branches will follow
master's
coding style and deviate as little (realistically) as possible from it.

Branches

master
: Standard GAN.
ac-gan
: Auxiliary classifier GAN as described in: Conditional image synthesis with auxiliary classifier GANs.
info-gan
: Information maximizing GAN as described in: InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets.
cGAN
: As described in: Image-to-Image Translation with Conditional Adversarial Networks.
wGAN
: As described in: Wasserstein GAN with improvements as described in: Improved Training of Wasserstein GANs.

SimGAN
here: https://github.com/wayaai/SimGAN.

Note:

ACGAN
is a more limited form of
InfoGAN
.
InfoGAN
can take an arbitrary number of categorical and continuous latent variables as input to the generator. ACGAN is an InfoGAN in the case where the generator takes one categorical latent variable as input corresponding to the label of the image to be generated.

wGAN
objective function should be used for all variations of GANs instead of the Jenson-Shannon divergence.

Notes

This repo and its branches were derived from Waya.ai's code base and are released in a cleaner and more modular form. I haven't fully tested each branch yet though so there may be some issues, and the GANs may need to be tuned a bit to converge properly.

About Waya.ai

Waya.ai is a company whose vision is a world where medical conditions are addressed early on, in their infancy. This approach will shift the health-care industry from a constant fire-fight against symptoms to a preventative approach where root causes are addressed and fixed. Our first step to realize this vision is easy, accurate and available diagnosis. Our current focus is concussion diagnosis, recovery tracking & brain health monitoring. Please get in contact with me if this resonates with you!

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.