Skip to content

Inception score for measuring quality of images generating from GAN

Notifications You must be signed in to change notification settings

nnUyi/Inception-Score

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Inception-Score

  • Inception score script for measuring quality of images generating from GAN

Method

  • This repo derived from paper named Improved Techniques for Training GANs, In this paper, we can see more tricks and improved methods used for training GAN. Anyway, sometimes, they are helpful.

  • Inception score is used to measure the quality of images generating from GAN model. It is considered as a good way to measure performance of GAN in quantity.

Requirements

  • tensorflow 1.3.0

  • python 2.7.12

  • numpy 1.13.1

  • scipy 0.17.0

Usages

download repo

$ git clone https://github.com/nnuyi/Inception-Score.git
$ cd Inception-Score

load data

  • In this repo, I provide 30 images sampled from cifar10. All the images are stored in the data directory(named data).

  • If you want to use your own data, images should be stored in data directory. And then launch the code.

launch repo

$ python inception_score.py

Results

  • In this repo, you can easily get the result below:

    cifar10 images cifar10 images
    Alt test Alt test
    inception score:
    mean:2.1532264
    stddev:0.27045175
    inception score:
    mean:2.1532264
    stddev:0.27045175

References

Contacts

About

Inception score for measuring quality of images generating from GAN

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages