jleinonen / keras-fid

Frechet Inception Distance for Keras-based GANs

Frechet Inception Distance for Keras GANs

This module contains an implementation of the Frechet Inception Distance (FID) metric for Keras-based generative adversarial network (GAN) generators.

The FID is defined here: https://arxiv.org/abs/1706.08500

Usage

A basic example:

import fid

generator = ... # Your code for creating the GAN generator
noise = ... # Your generator inputs (usually random noise)
real_images = ... # Your training dataset

# change (0,1) to the range of values in your dataset
fd = fid.FrechetInceptionDistance(generator, (0,1)) 
gan_fid = fd(real_images, noise)

If you already have the means and covariances:

gan_fid = fid.frechet_distance(mean1, cov1, mean2, cov2)

More information

See the docstrings for the fid.FrechetInceptionDistance class and the fid.frechet_distance function for more detailed documentation.