基于改进YOLOv5的绝缘子检测算法研究  

Research on insulator detection algorithm based on improved YOLOv5

作　　者：丁杰 蒋作[2] DING Jie;JIANG Zuo(School of Electrical and Information Engineering,Yunnan Minzu University,Kunming 650500,China;School of Mathematics and Computer Science,Yunnan Minzu University,Kunming 650500,China)

机构地区：[1]云南民族大学电气信息工程学院,云南昆明650500 [2]云南民族大学数学与计算机科学学院,云南昆明650500

出　　处：《云南民族大学学报（自然科学版）》2024年第4期505-512,共8页Journal of Yunnan Minzu University:Natural Sciences Edition

基　　金：国家自然科学基金(61866040).

摘　　要：针对绝缘子缺陷部分识别精度低问题,提出一种改进的YOLOv5算法,在原模型的特征融合部分引入动态卷积模块ODConv(omni-dimensional dynamic convolution),其通过并行多维注意力机制策略,增强模型对目标的特征提取能力.实验结果表明,改进后的算法比原算法召回率提高2.2%,精度提高3%,平均精度(MAP)提高2%,在NVIDIA GeForce RTX 30606G显存设备上速度达到172帧/s,对比多种主流目标检测算法,本文算法综合性能较优,可为输电线路绝缘子故障巡检提供技术参考.Aiming at the low accuracy of insulator defect recognition,this paper proposes an improved YOLOv5 algorithm,which introduces the dynamic convolution module ODConv(Omni-Dimensional Dynamic Convolution)in the feature fusion part of the original model.It enhances the feature extraction ability of the model to the target through the parallel multi-dimensional attention mechanism strategy.The experimental results show that the improved algorithm improves the recall rate by 2.2%and the accuracy by 3%compared with the original algorithm,The average accuracy(MAP)is improved by 2%,and the speed reaches 172 frames/s on NVIDIA GeForce RTX 30606G video memory device.Compared with many mainstream target detection algorithms,the algorithm in this paper has better comprehensive performance,which can provide technical reference for insulator fault patrol inspection of transmission lines.

关 键 词：YOLOV5 动态卷积 绝缘子缺陷检测 目标检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TM75[自动化与计算机技术—计算机科学与技术]