Are your reproduced results different from those in the author's paper?

ZhangXH227 / ftcn-training

1 stars 1 forks source link

Are your reproduced results different from those in the author's paper? #1

Open raining-dev opened 1 year ago

raining-dev commented 1 year ago

What is the test result of your recurring model on the dataset? AUC(%)：Celeb-DF-v2(CDF)、Deepfake Detection Challenge Preview(DFDC)、FaceShifter(FSh) and DeeperForensics(DFo)

WenOOI commented 7 months ago

@ZhangXH227 @is-jianjian-a @raining-dev Hi, did you reproduce the result? I carefully reproduced the code as mentioned in FTCN(paper: Exploring Temporal Coherence for More General Video Face Forgery Detection). But it overfits severely.