Survey on Generalization Methods of Face Forgery Detection

The rapid development of deep learning technology provides a powerful tool for deep forgery research. It is increasingly difficult for the human eye to distinguish the real and fake video images. Forged video images will have a huge negative impact on social life, such as financial fraud., fake news dissemination, personal bullying, etc. At present, deep learning-based fake face detection technology has achieved high accuracy on multiple benchmark databases (such as FaceForensics++), but the detection accuracy across databases is much lower than the source The detection accuracy in the database, that is, many detection methods are difficult to generalize to different or unknown types of forgery. Focus on the generalization research of face forgery detection methods based on deep learning, first of all, a brief introduction and comparison of commonly used databases for forgery detection are made. ;Secondly, the generalization of video image forgery detection methods is classified, summarized and analyzed from