基于PaddleHub的人脸口罩识别系统

A Face Mask Recognition System Based on PaddleHub

新冠病毒的传播方式主要是通过呼吸道传染,人们佩戴口罩进行防护是最有效的预防感染的途径之一,而人脸口罩自动化识别可以有效监督人们是否佩戴口罩。文章设计了基于PaddleHub的人脸口罩识别系统,利用PyramidBox单阶段人脸检测器和FaceBoxes主干网络分别对目标区域进行人脸检测和口罩识别。此外,为扩展该系统的功能,还增加了嵌入式人体测温和视频监控识别的核心功能,以及增加了分类人数统计、实时语音报警和实时截图的特色功能。该系统采用了网页端的形式进行了各功能的效果展示,并且在全天候自然场景下进行了实地测试达到了显著的检测效果。

Novel coronavirus is transmitted mainly through respiratory tract.Wearing masks for protection is one of the most effective ways to prevent infection,and automatic face mask recognition can effectively monitor whether people wear masks or not.The system used PyramidBox unit face detector and FaceBoxes network to detect the face and mask in the target area respectively.In addition,in order to expand the function of the systen,the system had also added core functions about embedded human body measurement,mild video surveillance and identification,as well as features of classified number statistics,real-time voice alarm and real-time screenshots.In this paper,the effects of each function were displayed in the form of web terminal,and the significant detection effect was achieved through field test under all-weather natural scene.