A new paper from researchers in Italy and Germany proposes a method to detect deepfake videos based on biometric face and voice behavior, rather than artifacts created by face synthesis systems, expensive watermarking solutions, or other more unwieldy approaches. The framework requires an input o
Definitely an interesting approach to this current arms race.
something i’ve wondered is why do we need to rely on such detailed methods? surely at a pixel distribution level there’s enough tells? why is that not the case?