Generalized face forgery detection with self-supervised face geometry information analysis network

Rui Zhang, Hongxia Wang, Hanqing Liu, Yang Zhou, Qiang Zeng

Abstract: Highlights•Enhance face forgery detection by leveraging information theory.•Develop a self-supervised method for identification of informative regions.•Apply Graph Convolutional Networks to analyze geometric relationships in facial landmarks and regions.•Evaluate the performance of the proposed methods on eight large-scale benchmark datasets.