大理大学学报 ›› 2022, Vol. 7 ›› Issue (12): 37-42.

• 物理学 • 上一篇    下一篇

基于YOLOv5s的施工人员安全防护设备佩戴检测研究

  

  1. 1.大理大学工程学院,云南大理 6710032.浙江工商大学信息与电子工程学院,杭州 310018

  • 收稿日期:2022-05-05 出版日期:2022-12-15 发布日期:2022-12-15
  • 作者简介:赵恩铭,副教授,博士,主要从事目标检测、嵌入式研究。
  • 基金资助:
    国家自然科学基金项目(62065001);云南省中青年学术和技术带头人后备人才项目(202205AC160001);教育部2020年产学合作协同育人项目(202002285009);云南省地方本科高校(部分)基础研究联合专项资金项目(2019FH001-066);浙江省自然科学基金基础公益研究计划项目(LGG20E040001)

Research on Construction Workers' Safety Protection Equipment Wearing Detection Based on YOLOv5s

  1. 1.College of EngineeringDali UniversityDaliYunnan 671003China2.School of Information and Electronic EngineeringZhejiang Gongshang UniversityHangzhou 310018China

  • Received:2022-05-05 Online:2022-12-15 Published:2022-12-15

摘要: 针对施工人员作业期间存在不规范佩戴安全帽、口罩的问题,采用YOLOv5s算法对施工人员进行安全防护设备佩戴检测,降低安全事故发生的概率。YOLOv5s算法与嵌入式平台Jetson Nano结合,使用TensorRT技术提升YOLOv5s算法的检测帧率,最终模型的mAP值为0.946,在Jetson Nano上的检测帧率为12.83 FPS。将YOLOv5s算法应用于施工现场安全保障领域,实现了对施工人员安全帽、口罩佩戴情况的有效监测,满足实际应用中检测速度与设备便携安装的需求。

关键词:

YOLOv5s, 目标检测, TensorRT, Jetson Nano

Abstract:

To solve the problem of irregular wearing of safety helmets and masks by construction workers' during operation the YOLOv5s algorithm is used to detect the wearing of safety protection equipment for construction workers to reduce the probability of safety accidents. The YOLOv5s algorithm is combined with the embedded platform Jetson Nano and the TensorRT technology is used to improve the detection frame rate of the YOLOv5s algorithm. The mAP value of the final model is 0.946 and the detection frame rate on the Jetson Nano is 12.83 FPS. The YOLOv5s algorithm is applied to the field of construction site safety assurance which realizes effective monitoring of the wearing of construction workers' helmets and masks and meets the needs of detection speed and portable installation of equipment in practical applications.

Key words: YOLOv5s, target detection, TensorRT, Jetson Nano

中图分类号: