A face detection method, face detection model training method and device

Abstract

This application discloses a face detection method for protecting user privacy and improving information security, which can be applied to the field of artificial intelligence. This application includes: obtaining a sequence of human face images; obtaining a sequence of noise images according to the sequence of human face images; The first feature extraction network of the face detection model obtains the face feature map sequence of the face image sequence; the noise feature map sequence of the noise image sequence is obtained through the second feature extraction network of the face detection model; based on the face feature map sequence and Noise feature map sequence, the classification probability value corresponding to the face image is obtained through the fully connected layer of the face detection model; the detection result is determined according to the classification probability value. The present application also provides a training method and device for a face detection model. The present application pays attention to the semantic information of the face image sequence itself and the noise information of the noise image sequence at the same time, so as to effectively discover artifacts produced by forged faces and improve the accuracy of face detection.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, in particular to a face detection method, a face detection model training method and a device. Background technique [0002] With the development of artificial intelligence (AI) technologies such as generative confrontation networks, deep face generation technology and its applications are becoming more and more mature. People can quickly realize face generation, face editing and face replacement through neural networks. Face generation technology has promoted the emerging development of the entertainment and cultural exchange industries, but it also poses a huge potential threat to face security. [0003] At present, a face detection model based on forgery traces of faces is provided, and the face detection model mainly judges specific forgery traces of false content. Because there will be unnatural blinking in the generated face-changing images or videos, the face det...

AI Technical Summary

Problems solved by technology

[0004] However, face detection models based on face forgery traces only perform well for face-swapped videos containing specific forgery traces, but lack generalization

If the training of blinking face images is added to the face forgery algorithm, the generated face-changing images or videos will be closer to the natural blinking phenomenon, resulting in lower detection accuracy of the face detection model

Images