Skip to content

matthew-liu/beta-vae

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Face Generation Using Variational Autoencoders

This repo contains training code for two different VAEs implemented with Pytorch.
I used the CelebA Dataset for training, with 182637 training images and 19962 testing images.
Trained model can be found in /checkpoints.

poster

Model structures:

beta-vae

dfc-vae

Results after 300 epochs:

Original Faces (Top) vs. Reconstructed (Bottom) :

r1 r2 r3 r4

Linear Interpolation from z1 (leftmost) to z2 (rightmost):

l1

Vector Arithmetic from original (leftmost) to wearing sunglasses (rightmost):

l1

Generated Images with randomly sampled latent z ~ N(0, 1):

l1

(Notes: output images above are results after image dehazing using this script [3])

References

[1] β-VAE: https://openreview.net/pdf?id=Sy2fzU9gl
[2] DFC-VAE: https://arxiv.org/abs/1610.00291
[3] Dehaze: https://github.com/cssartori/image-dehazing.git

About

A Pytorch Implementation of the Beta-VAE

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages