摘 要:针对后疫情时代人脸识别系统无法在佩戴口罩的情况下准确识别人脸的问题,基于 OpenCV 库和 Dilb 库架构系统,引入 YOLOv5 目标检测算法替代 Dlib 原有的人脸目标区域检测算法。通过检测算法获取一系列目标特征值的坐标位置后,采用Centriod Tracking 目标追踪算法来处理这些特征,从而实现对佩戴口罩人脸图像的快速识别,系统同时实现了用户注册、数据录入、信息管理和日志等功能。
关键词:深度学习;YOLOv5 算法;目标追踪算法;戴口罩人脸识别
DOI:10.19850/j.cnki.2096-4706.2023.06.016
基金项目:2022 年西安工业大学大学生创新创业训练计划项目(S202210702109)
中图分类号:TP391.4 文献标识码:A 文章编号:2096-4706(2023)06-0061-04
Face with Mask Recognition System Based on YOLOv5 and Target Tracking Algorithm
WU Jingwen, LI Xiaolong, LIANG Xiangyang
(School of Computer Science and Engineering, Xi'an Technological University, Xi'an 710021, China)
Abstract: In view of the problem that the face recognition system cannot accurately recognize faces under the condition of wearing masks in the post-epidemic era, based on the OpenCV library and Dilb library architecture system, the YOLOv5 object detection algorithm is introduced to replace the original face target area detection algorithm of Dlib. After the coordinate position of a series of target feature values is obtained by the detection algorithm, the Centriod Tracking target tracking algorithm is used to process these features, so as to realize the rapid recognition of face images with masks. The system also realizes the functions of user registration, data entry, information management and log.
Keywords: deep learning; YOLOv5 algorithm; target tracking algorithm; face with mask recognition
参考文献:
[1] 赵畅 . 基于 YOLOv5 改进的人脸检测算法的研究与实现[D]. 长春:吉林大学,2022.
[2] 王鑫,王忠举,李锐 . 基于神经网络的人脸识别研究综述[J]. 信息与电脑:理论版,2020,32(23):56-58.
[3] 赵元章 . 人脸遮挡物识别算法的研究与应用 [D]. 西宁:青海师范大学,2022.
[4] 黄渶柯 . 基于深度学习的口罩人脸识别系统研究 [D]. 长沙:湖南大学,2021.
[5] 黄振龙,吴林煌.基于Yolov5s和Dlib的视频人脸识别 [J].电脑知识与技术,2021,17(32):94-96.
[6] 肖博健,万烂军,陈俊权 . 采用 YOLOV5 模型的口罩佩戴识别研究 [J]. 福建电脑,2021,37(3):35-37.
作者简介:吴静雯(2002—),女,满族,陕西西安人,本科在读,研究方向:智能科学与技术;李小龙(2001—),男,汉族,陕西渭南人,本科在读,研究方向:软件工程。