Hi,
Regarding your repository's description, you have used faceforencis++ for training. So I wanna ask 2 questions:
first, this dataset has more than 1.8 million fake images, would you please tell me which videos (deepFakes, face2face, faceSwap, neuralTextures) you have used as the fake part of your 0.96 million dataset?!
second, when a network trains based on the manipulated faces, does it able to recognize printed attacks or replay attacks?! I mean it seems that the nature of this dataset is different from our object. Am I make a mistake?!
Thank you in advance for your response.
Hi,
Regarding your repository's description, you have used faceforencis++ for training. So I wanna ask 2 questions:
first, this dataset has more than 1.8 million fake images, would you please tell me which videos (deepFakes, face2face, faceSwap, neuralTextures) you have used as the fake part of your 0.96 million dataset?!
second, when a network trains based on the manipulated faces, does it able to recognize printed attacks or replay attacks?! I mean it seems that the nature of this dataset is different from our object. Am I make a mistake?!
Thank you in advance for your response.