I was wondering if the Frechet inception distance for two colored datasets would be the same than the FID calculated for the same datasets converted to grayscale.

I know that it depends on the feature extraction, which is the Inception network. And that is the question, I don't fully understand the role of color in CNNs. I thought that all color information is lost in the feature extraction, and in this case I grayscale and colored datasets would produce the same FID. But I am not sure about that.

  • $\begingroup$ Can you please provide us with a link to an article that talks about this "Frechet Inception Distance". I know it exists, but not all people may know what exactly you're referring to. Moreover, it may be a good idea to tell us what you know about the FID. What is it usually used for? What's the equation to compute it? Generally, it's a good idea to do a little bit of research before asking a question. $\endgroup$
    – nbro
    Nov 20 '21 at 21:43
  • 1
    $\begingroup$ Hi, I added more info $\endgroup$
    – Manveru
    Nov 21 '21 at 15:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.