It would be interesting to provide metrics for generated image evaluation like here:
@vfdev-5
I am interested in this issue.
I think that possible metrics would be inception scores, FID, PPL, or LPIPS.
Do you have any priorities? I can start working on one of them.
@kamahori thanks ! You can start by any metric you'd like. Maybe, those for stylegan2 would be a little bit more prioritized.
@vfdev-5 Okay, then let me start with FID.