Identifying Celebrity Deepfakes From Outer Face Regions

A new collaboration between Microsoft and a Chinese university has proposed a novel way of identifying celebrity deepfakes, by leveraging the shortcomings of current deepfake techniques to recognize identities that have been'projected' onto other people. The approach is called Identity Consistency Transformer (ICT), and works by comparing the outermost parts of the face (jaw, cheekbones, hairline, and other outer marginal lineaments) to the interior of the face. The system exploits commonly available public image data of famous people, which limits its effectiveness to popular celebrities, whose images are available in high numbers in widely available computer vision datasets, and on the internet. The forgery coverage of faked faces across seven techniques: DeepFake in FF; DeepFake in Google DeepFake Detection; DeepFaceLab; Face2Face; FSGAN; and DF-VAE. Popular packages such as DeepFaceLab and FaceSwap provide similarly constrained coverage.