search

tamarott / SinGAN

Official pytorch implementation of the paper: "SinGAN: Learning a Generative Model from a Single Natural Image"
https://tamarott.github.io/SinGAN.htm
Other
3.31k stars 611 forks source link

Why the image format have so much impact of SIFID, could you give some explanation? #124

Open lioo717 opened 4 years ago

lioo717 commented 4 years ago

Thank you for noticing this. The format should be jpg. Code and readme file were updated.

_Originally posted by @tamarott in https://github.com/tamarott/SinGAN/issue_comments/597523136_

Why the image format have so much impact of SIFID. It seems weird, and I can't understand. Could you give some explanations on it?

A-chen23 commented 4 years ago

Excuse, I used the same format .jpg. python SIFID/sifid_score.py --path2real "Input/Images/birds.jpg" --path2fake "fake_image/birds.jpg" But the same error is prompted (SIFID: nan). Can you give me some advice? thank you

lioo717 commented 4 years ago

Excuse, I used the same format .jpg. python SIFID/sifid_score.py --path2real "Input/Images/birds.jpg" --path2fake "fake_image/birds.jpg" But the same error is prompted (SIFID: nan). Can you give me some advice? thank you

Sorry, I never have this problem. Maybe you should check overflow and underflow in the code. Just a conjecture.