0
$\begingroup$

If I want to measure the quality of the images generated from GAN, what metric should I use and any reference for it so that I can check how it's done?

I was thinking of using Frechet Inception Distance (FID) score, but had one doubt. Is FID score suitable if we want to measure the quality of individual images? Or, FID only used to compare distribution of n real images vs n generated images?

$\endgroup$

1 Answer 1

0
$\begingroup$

The FID score compares the distribution of generated images with the distribution of a set of real images, providing insights into the realism and diversity of the generated images. It calculates the distance between feature vectors of real and generated images, summarizing how similar the two groups are in terms of statistics on computer vision features of the raw images. Lower FID scores indicate higher quality and diversity in the generated images, with a perfect score of 0.0 indicating that the two groups of images are identical.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.