Skip to content

PyTorch-Ignite compatible implementation of a GAN metric called WInD

Notifications You must be signed in to change notification settings

aalbayrak394/ignite-wind

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 

Repository files navigation

ignite-wind

A PyTorch-Ignite compatible implementation of the Wasserstein Inception Distance (WInD) introduced in https://ieeexplore.ieee.org/document/9053325 for evaluating GANs inspired by https://github.com/streetakos/wasserstein-distance.

WInD extends the assumption in state of the art evaluation measure FID of the data distributions being Gaussians and fits Gaussian Mixture Models on the inception features of real and fake data distribution respectively. More details can be read in the original paper.

About

PyTorch-Ignite compatible implementation of a GAN metric called WInD

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages