基于YOLOV7的井下机车运行中潜在碰撞体检测方法  

Potential collision detection method for underground locomotive operation based on YOLOV7

作　　者：刘伟新 张洪昌 任宇昕 何江 王迎镇 LIU Wei-xin;ZHANG Hong-chang;REN Yu-xin;HE Jiang;WANG Ying-zhen(Ansteel Group Mining Co.,Ltd,Anshan 114001,China;School of Resources and Civil Engineering,Northeastern University,Shenyang 110819,China;Beijing University of Science and Technology Civil and Resource Engineering,Beijing 100083,China)

机构地区：[1]鞍钢集团矿业有限公司,辽宁鞍山114001 [2]东北大学资源与土木工程学院,辽宁沈阳110819 [3]北京科技大学,北京100083

出　　处：《世界有色金属》2024年第21期22-24,共3页World Nonferrous Metals

摘　　要：面对地下矿山无人驾驶电机车运行过程中的复杂环境,如何高效、迅速、精准的识别出机车前进方向上的人员、矿石等潜在碰撞体,对提升矿山运输效率、增长矿山经济效益具有重要意义。Yolov7算法作为yolo系列较新且应用十分广泛的算法,具有响应速度快、准确率高、运行稳定等特点,因此本文尝试使用yolov7算法来进行潜在碰撞体检测实验。试验结果表明,yolov7算法精确度符合要求,mAP达到96.5%,且在低光照、远距离等环境下均可对潜在碰撞体进行精准检测。In the face of the complex environment during the operation of the underground mine unmanned electric locomotive,how to efficiently,quickly and accurately identify the potential collision bodies such as personnel and ore in the direction of the locomotive is of great significance to improve the efficiency of mine transportation and increase the economic benefits of the mine.As a relatively new and widely used algorithm of yolo series,the yolov7 algorithm has the characteristics of fast response speed,high accuracy,stable operation,etc.Therefore,this paper attempts to use the yolov7 algorithm to conduct potential collision body detection experiments.The test results show that the accuracy of the yolov7 algorithm meets the requirements,with mAP reaching 96.5%,and it can accurately detect potential colliders in low-light and long-distance environments.

关 键 词：目标检测 地下矿山 yolov7 深度学习 

分 类 号：TD64[矿业工程—矿山机电]