5G物联网框架下安全帽的深度学习识别方法  

Deep learning recognition method for safety helmet in the framework of 5G IoT

作　　者：方琪琦 熊引 代阳 FANG Qi-qi;XIONG Yin;DAI Yang(CHN ENERGY Railway Equipment Co.,Ltd.,Beijing 100010,China;Shanghai Rentong Electronic Technology Co.,Ltd.,Shanghai 200233,China;School of Mechanical and Electrical Engineering,Xi’an Polytechnic University,Xi’an 710600,China)

机构地区：[1]国能铁路装备有限责任公司,北京100010 [2]上海仁童电子科技有限公司,上海200233 [3]西安工程大学机电工程学院,西安710600

出　　处：《信息技术》2025年第4期46-53,60,共9页Information Technology

基　　金：陕西省科技厅自然科学基础研究计划-面上项目(2022JM-219)。

摘　　要：针对目前安全帽检测存在远程传输存在延迟、数据接口和格式不统一、难以实现跨区域的监管的问题,文中提出了一种基于5G和物联网技术的安全帽检测方法。该方法针对不同监测区域通过部署网络摄像头和5G网关,实现实时视频数据流的远程传输,数据流经过本地的边缘服务器进行预处理。边缘服务器将形成统一视频监控系统接口,然后对接物联网平台。通过部署在服务器的YOLOv5深度学习算法实现对安全帽的检测。采用基于Flask的Web服务框架,以实现客户端请求的处理和响应。系统检测速率为60fps,精度为98.5%,可以应用在大规模集成化的安全管控系统中。A safety helmet detection method based on 5G and Internet of Things(IoT)technology is proposed to address the current issues of delay in remote transmission,inconsistent data interfaces and formats,and difficulties in achieving cross-regional supervision.Network cameras and 5G gateways are deployed in different monitoring areas to enable real-time video data streaming for remote transmission.The data stream undergoes preprocessing by local edge servers.These edge servers form a unified video surveillance system interface,which is then connected to the IoT platform.The YOLOv5 deep learning algorithm deployed on the server is used for safety helmet detection.The web service framework of Flask is employed to handle client requests and responses.The system achieves a detection rate of 60 frames per second(fps)with an accuracy of 98.5%and can be applied in large-scale integrated safety control systems.

关 键 词：安全帽检测 深度学习 YOLOv5 5G物联网 FLASK 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]