Method for generating front face picture from side face picture based on generative adversarial network

Abstract

The invention discloses a method for generating a front face picture from a side face picture based on a generative adversarial network, and the method comprises the steps of enabling collected side face pictures and front face pictures to be used for training the generative adversarial network, then collecting the side face pictures again, inputting the side face pictures into the trained generative adversarial network, and outputting the front face picture. According to the method for generating the front face picture, existence of local details of the face is guaranteed through a double-path generator, effective feature extraction of a contour sub-generator is accelerated by predicting the positions of sense organs and contours, and it is guaranteed that the two features can act and result through a splicing mode; therefore, the high-quality image which can be used for face recognition is generated. According to the method, the local features and the global features are reserved, and the generated front face picture has the characteristics of accuracy and high efficiency.

Description

technical field [0001] The present invention relates to the technical field of human face generation, in particular to a method for generating a frontal face image from a profile face image based on a generative confrontation network. Background technique [0002] Current face recognition systems and face databases use frontal face pictures as standard data, so the face recognition model after training has a good performance on the recognition of frontal faces, but the performance is extremely poor for data whose input is profile faces. Difference. In real life, most of the data captured by the camera is profile data, which cannot be directly applied to the face recognition system. Especially for security field or smart home field. For the former, a large number of cameras will be deployed in public places to search and record criminals, and criminals will deliberately avoid the cameras, so that the cameras cannot capture or only capture his side face, while the face recog...

AI Technical Summary

Problems solved by technology

[0005] 1. The method based on feature extraction cannot be applied to mature face recognition modules;

[0006] 2. The method based on feature generation will cause the lack of details or the invalidity of features

Images