Application of face recognition in port unrestricted scene
Song Qi, Hanbing Yao, WuWenBo · 2021
Traditional face recognition technology needs to collect data in a specified environment with stable facial features. However, due to the large scope and wide area of port operation, traditional identification methods have been unable to meet the requirements of port unrestricted scenarios. In this paper, the algorithms based on RetinaFace face detection and ArcFace face recognition are applied to port staff attendance in unrestricted scenarios. In order to obtain a more stable and accurate face representation method, this paper studies face detection, face image pretreatment, face recognition, loss function and other aspects, and verifies the feasibility of this model in port unrestricted scene attendance through the experimental results on PubFig dataset and the analysis of real-time face recognition effect.