Face recognition method based on combining with face attribute information

Abstract

The invention discloses a face recognition method based on combining with face attribute information and belongs to the technical field of digital image processing. The invention aims at overcoming the technical problems that the existing integration method requires training of multiple DCNN networks and then fractional fusion or feature fusion for further training, has tedious and complicated tasks and is not conducive to practical applications, and discloses a novel way of integrating identity information and attribute information to improve the accuracy of face recognition. The face recognition method integrates a face identity authentication network and an attribute recognition network to form an integration network and adopts a joint learning method to simultaneously learn identity characteristics and face attribute characteristics, thereby not only improving the accuracy of face recognition, but also enabling prediction of human faces; the integration network is a multitasking network; besides, the face recognition method adopts a cost-sensitive weighting function so that it does not depend on target domain data distribution, but realizes balanced training in source data domain; and the modified integration framework only adds a few parameters and the extra computational load is small.

Description

technical field [0001] The invention belongs to the technical field of digital image processing, and in particular relates to a face recognition method based on combining face attribute information. Background technique [0002] With the rapid development of deep learning, face recognition technology has developed by leaps and bounds, and many products using face recognition technology have emerged as the times require. However, the current face recognition technology still has many limitations. Among them, the very typical problems are the influence of environmental factors such as large side faces and lighting, which will reduce the performance of the face recognition system. Many researchers have done a lot of work on face pose correction and domain adaptation. Although they have achieved many results, they are still in the exploratory stage. After research, it is found that many face attribute information (such as: gender, eyebrow shape, nose bridge height, etc.) are no...

AI Technical Summary

Problems solved by technology

[0007] The above two methods need to train multiple DCNN networks, and then perform score fusion or feature fusion for further training. The tasks are heavy and complicated, which is not conducive to practical application.

Images