Skip to content

Official implementation of "Learning to Discover Cross-Domain Relations with Generative Adversarial Networks"

Notifications You must be signed in to change notification settings

Tanvi17/semantic-face-editing

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

28 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DiscoGAN

Prerequisites

  • Python 3.8
  • PyTorch
  • Numpy/Scipy/Pandas
  • Progressbar
  • OpenCV

Training DiscoGAN

Download CelebA dataset using:

$ python ./datasets/download.py celebA 

(Currently, the link for downloading CelebA dataset is not available).

To train gender conversion:

$ python ./discogan/image_translation.py --task_name='celebA' --style_A='Male'

To train hair color conversion:

$ python ./discogan/image_translation.py --task_name='celebA' --style_A='Blond_Hair' --style_B='Black_Hair' --constraint='Male'

Results

Example result shows images generate from x_A, x_AB, x_ABA and x_B, x_BA, x_BAB models

Eyeglasses

Hair color conversion

Gratitude

This project would not be possible without the mentorship of Prof. J. Kao and TAs of course ECE 247: Neural Networks and Deep Learning. A huge shoutout to SKT-Brain for DiscoGAN. Checkout the reference here: https://github.com/SKTBrain/DiscoGAN

About

Official implementation of "Learning to Discover Cross-Domain Relations with Generative Adversarial Networks"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 96.4%
  • Shell 3.6%