Face depth forgery detection method and system based on face division

Abstract

The invention provides a face depth forgery detection method and system based on face division. The method comprises the following steps: extracting global face features from training data; according to shallow convolution features generated in the process of obtaining the global face features, dividing the shallow convolution features into a plurality of image areas according to a preset face division mode, and inputting the image areas into a local face feature extraction model to obtain a plurality of local features of a face image; and extracting relation features among the plurality of local features through an attention model, splicing the relation features and the global features, inputting the spliced relation features and global features into a dichotomy model to obtain a detection result of the training data, and constructing a loss function according to the result and a label to train a global face feature extraction model, the local face feature extraction model, the attention model and the dichotomy model.

Description

technical field [0001] The method belongs to the field of machine learning and computer vision, and in particular relates to machine learning problems for deepfake detection in computer vision. Background technique [0002] In recent years, deep learning technology has been continuously developed and has been widely used in the field of computer vision. On the one hand, deep learning technology has led a new wave of artificial intelligence, but on the other hand, a series of security issues caused by deep learning have also attracted more and more attention. At present, image and video recognition technologies based on deep learning are widely used in all aspects of people's lives, such as intelligent supervision of network content, automatic video monitoring and analysis, access control systems based on face recognition, and facial recognition payment. In these key application fields, the reliability and security of information and data should be paid attention to and guar...

AI Technical Summary

Problems solved by technology

But the disadvantage is that in the single-frame approach, there are few methods that can adaptively focus on local forged areas and find suspicious areas.

[0010] At present, in the field of deep forgery detection, methods for face segmentation have also been proposed, but the existing methods have the following problems: (1) These methods divide the face in the early stage of the model, that is, the image input to the model is segmented, and the model only A small part of the face image can be seen, and some features are not shared for global information and local information

(2) Existing models only use this part of features for classification after acquiring regional features, and there is no good way to model the relationship between regions

Images