search

yyk-wew / F3Net

Pytorch implementation of F3Net (ECCV 2020 F3Net: Frequency in Face Forgery Network)
148 stars 17 forks source link

about dataset #19

Open ss880426 opened 2 years ago

ss880426 commented 2 years ago

Excuse me, does the author use the LQ dataset to train and then test LQ or the HQ dataset to train and then test LQ? thanks!!!

yyk-wew commented 2 years ago

I think the experiments are all done under the same quality setting, i.e., train on LQ and test on LQ.