基于改进YOLOv8的钻机目标检测模型  

Target Detection Model of Drilling Rig Based on Improved YOLOv8

作　　者：李腾飞 芦碧波[1] 李小军[2] Li Tengfei;Lu Bibo;Li Xiaojun(School of Computer Science and Technology,Henan Polytechnic University,Jiaozuo 454000,China;School of Energy Science and Engineering,Henan Polytechnic University,Jiaozuo 454000,China)

机构地区：[1]河南理工大学计算机科学与技术学院,河南焦作454000 [2]河南理工大学能源科学与工程学院,河南焦作454000

出　　处：《煤矿机械》2025年第3期195-197,共3页Coal Mine Machinery

基　　金：河南省高等学校重点科研项目(24B520013);河南省重点研发与推广专项(科技攻关)项目(222102210131);河南理工大学基本科研业务费专项项目(自然科学类)(NSFRF240508)。

摘　　要：由于井下环境复杂,低光照、人员遮挡、眩光等多种因素都会影响钻机检测精度。针对上述问题,提出一种钻机目标检测模型CAS-YOLOv8。为减少低光照因素干扰,设计了C2fGE结构,增强不同通道之间的信息融合;为解决人员遮挡问题,提出ADownCS模块,提高特征提取能力;对于眩光问题,在网络中增加混洗注意力;为减少环境中小目标对检测的干扰,依据钻机尺寸,去除特征图较小的检测头。试验结果表明,与原始YOLOv8相比,CAS-YOLOv8的mAP@0.5增加4.8%、参数量降低73.8%。Due to the complexity of the underground environment,a variety of factors such as low light,people occlusion and glare can affect the detection accuracy of drilling rig.Aiming at the above problems,a target detection method CAS-YOLOv8 of drilling rig was proposed.In order to reduce the interference of low-light factors,the C2fGE structure was designed to enhance the information fusion between different channels.To solve the problem of people occlusion,the ADownCS module was proposed to improve the feature extraction capability.For the problem of glare,the shuffle attention was added to the network.In order to minimize the interference of small targets in the environment,detectors with small feature maps were removed based on the size of drilling rig.The experimental results show that compared with the original YOLOv8,the mAP@0.5 of CAS-YOLOv8 is improved by 4.8%and reduces the parameters by 73.8%.

关 键 词：钻机 视频分析 目标检测 YOLOv8 注意力 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]