基于改进YOLOv5s的轮毂气门孔检测算法  

Hub valve hole detection algorithm based on improved YOLOv5s

作　　者：刘锡琳 潘文松 张爱军[1] LIU Xilin;PAN Wensong;ZHANG Aijun(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China;Hangzhou KDx Biotechnology Co.,Ltd.,Hangzhou 311199,China)

机构地区：[1]南京理工大学机械工程学院,江苏南京210094 [2]杭州康代思锐生物科技有限公司,浙江杭州311199

出　　处：《电子设计工程》2024年第19期140-144,149,共6页Electronic Design Engineering

摘　　要：轮毂动平衡对车辆行驶安全尤为重要,气门孔快速的检测定位有利于提高动平衡测试效率。针对气门孔目标尺寸小、特征识别困难等问题,提出一种基于YOLOv5s的气门孔检测算法。该算法联合多尺度特征,扩大特征融合网络的输入尺寸范围,提高小目标的检测准确度;融合CBAM注意力机制,使网络更加关注关键区域,同时忽略不重要的背景信息;使用高效聚合网络,通过增加相当的深度来提高准确率。实验结果表明,改进后的算法在气门孔数据集上的识别精度达到92.64%,召回率达到84.16%。Hub dynamic balance is very important for the safety of vehicles.Rapid detection and positioning of valve holes is beneficial to improve the efficiency of dynamic balance test.Aiming at the problems of small target size and difficult feature recognition,a valve hole detection algorithm based on YOLOv5s was proposed.The algorithm combines multi-scale features,enlarges the input size range of feature fusion network,and improves the detection accuracy of small targets.The attention mechanism of CBAM is integrated to make the network pay more attention to key areas while ignoring unimportant background information.Use highly efficient aggregation networks to increase accuracy by adding considerable depth.The experimental results show that the improved algorithm has a recognition accuracy of 92.64%and a recall rate of 84.16%on the valve hole data set.

关 键 词：YOLOv5 气门孔 注意力机制 高效聚合网络 

分 类 号：TN29[电子电信—物理电子学]