摘要
煤矿井下员工的不安全行为是导致事故发生的主要原因,传统的矿工不安全行为防控主要依靠人来管控,智能化水平较低,难以实时自动发现并智能决策预警。文章将计算机视觉、深度学习相关技术结合,应用于煤矿井下员工不安全行为识别。基于YOLOv5目标检测算法、OpenPose人体姿态估计算法对视频数据中物的状态及人的行为进行分析,并提出一种行为判定方式,来识别矿工的不安全行为。将煤矿井下物的不安全状态及人的不安全行为进行结合,提出一种实时分析物的不安全状态和人的不安全行为的方法,有助于实现煤矿井下广泛场景中矿工不安全行为的自动识别,可用于应用层服务系统,实现预警等功能,为矿工不安全行为智能识别和预警提供新的思路和方法。
The unsafe behavior of coal mine staff is the main reason leading to the accident.The traditional prevention and control of the unsafe behavior of coal miners mainly depends on people to manage and control,and the level of intelligence is low,it is difficult to detect and intelligent decision-making early warning in real-time.In this paper,computer vision and deep learning techniques are combined to identify unsafe behaviors of coal mine employees.Based on YOLOv5 target detection algorithm and OpenPose pose estimation algorithm,the state of object and human behavior in video data are analyzed.Combining the unsafe state of underground objects and the unsafe behavior of human beings,a method of real-time analyzing the unsafe state of objects and the unsafe behavior of human beings is put forward,it can be used in the application layer service system to realize the functions of early warning and so on,it provides new ideas and methods for intelligent recognition and early warning of unsafe behavior of miners.
作者
侯艳文
姚有利
贾泽琳
苏丹
王晓义
郭康
HOU Yanwen;YAO Youli;JIA Zelin;SU Dan;WANG Xiaoyi;GUO Kang(College of Coal Engineering,Shanxi Datong University,Datong 037003,China)
出处
《煤》
2023年第11期33-36,91,共5页
Coal
基金
山西省软件科学研究项目(2019041014-3)
大同市科技项目:绿色开采背景下的矿工不安全行为智能识别方法研究项目(2023067)。
