Research of the occluded face recognition methods

Abstract

In face recognition, deep learning has always been a technology and problem that people need to improve. With the development of time, deep learning has the characteristics of self-learning ability, strong expression ability and better stability, and it is an effective solution to face recognition. However, in the sheltered environment, deep learning still faces many challenges. Different methods are used to expand the new deep learning methods, which can effectively solve the face recognition features in occlusion environment. This paper compares and classifies the related algorithms of deep learning and the traditional algorithms of deep learning. In particular, the network structure and construction principle of some classical algorithms are reviewed. Finally, discusses and summarizes their future development trends and directions. This study has certain reference significance for related scholars.