| Face recognition is the most popular biometric method because of its uniqueness,universality,practicality and easy access.It has received great attention in both industry and academia.However,when the face recognition system falls on the ground in real life,it often faces the situation of multiple faces.If the recognition object is required to recognize one by one,the efficiency is low.Currently,the research on face recognition algorithms for multi-face scenes is relatively few.It has great scientific and commercial value to propose a multi-face synchronous recognition algorithm suitable for large throughput of human flow.This paper analyzes the challenges faced by simultaneous recognition of multiple faces in real-world scenarios,such as the differences and occlusions of different face postures,more complex environments,and the decline of face speed in batch processing.To solve the above problems,this paper presents a multi-face synchronous recognition algorithm based on deep learning.First,a new designed feature extraction network is proposed,which combines new data enhancement methods and optimized detection process to accurately detect and align face regions in a multi-face environment.After that,a lightweight network structure with a faster speed and more facial features can be extracted,and the model size is only 6.7Mb,which is validated in multiple face test recognition datasets.Finally,a multi-face synchronization recognition system with improved algorithm is designed and developed,which has a friendly interface and is easy for users to use.The algorithm proposed in this paper improves the shortcomings of the current face recognition algorithm in the application of multi-face scenes,and solves the problem that the speed and accuracy of detection,alignment and recognition decrease due to the increase of the number of faces. |
