Skip to content

Likeness Score (LS) to evaluate GANs performance, based on direct analysis of generated images

License

Notifications You must be signed in to change notification settings

ShuyueG/GAN_evaluation_LS

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 

Repository files navigation

A Novel Evaluation to GANs

This is an example of computing the Likeness Score: LS on datasets of real images and generated images by GANs.

It includes codes to compute the 2-class distance-based separability index (DSI). LS=1-DSI. There are two versions (CPU and GPU).

Related paper

A Novel Measure to Evaluate Generative Adversarial Networks Based on Direct Analysis of Generated Images

Citation Guan, S., Loew, M. A novel measure to evaluate generative adversarial networks based on direct analysis of generated images. Neural Comput & Applic (2021). doi: 10.1007/s00521-021-06031-5.

Paper Arxiv

Time Complexity

Data: 2000 real images and 2000 generated images from CIFAR-10 dataset (32x32 RGB images); CPU: i7-6900K; GPU: GTX 1080Ti

  • CPU ver. 110.859 s
  • GPU ver. 2.406 s

About

Likeness Score (LS) to evaluate GANs performance, based on direct analysis of generated images

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages