基于YOLOv5s的轻量化轮胎缺陷检测算法研究  

Research on Lightweight Tire Defect Detection Algorithm Based on YOLOv5s

作　　者：廖明习 吴超 黄燕清 李卓 万林辰 韩一 LIAO Mingxi;WU Chao;HUANG Yanqing;LI Zhuo;WAN Linchen;HAN Yi(Undergraduate School,Wuhan University of Technology,Wuhan 430070,China;不详)

机构地区：[1]武汉理工大学本科生院,湖北武汉430070 [2]武汉理工大学信息工程学院,湖北武汉430070 [3]上汽通用五菱汽车股份有限公司,广西柳州545000

出　　处：《武汉理工大学学报(信息与管理工程版)》2025年第1期109-113,125,共6页Journal of Wuhan University of Technology:Information & Management Engineering

摘　　要：轮胎对车辆的性能、操控、安全性和乘坐舒适性都有着重要的影响,对轮胎缺陷进行检测能够及时发现问题,确保轮胎的安全性和可靠性。针对汽车胎侧起鼓、胎侧凹陷、胎侧开裂、外胎有小孔、胎侧气泡等多种轮胎缺陷检测,提出一种基于改进YOLOv5s的轮胎缺陷检测模型shuffle_yolov5s_cbam,引入置信度损失函数,将YOLOv5s中的主干网络替换为更轻量的ShuffleNetV2网络,将ShuffleNetV2网络中的激活函数ReLU替换为H-Swish,并引入CBAM注意力机制以增强模型对关键特征的关注。相比较于YOLOv5s模型,shuffle_yolov5s_cbam模型的平均mAP值提升了1.5%,检测速率缩短了17.3%,且能够更准确地识别和定位缺陷,不仅提高了产品质量和生产效率,还减少了人工检测的工作量和错误率。Tires have an important impact on vehicle performance,handling,safety and ride comfort.The detection of tire defects can help find problems in time and ensure the safety and reliability of tires.To detect various tire defects such as bulging on the tire sidewalls,indentations on tire sidewalls,cracks on tire sidewalls,pinholes in the outer tire,and air bubbles on tire sidewalls,this paper proposed a tire defect detection technology based on an improved YOLOv5s model.This technology introduced a confidence loss function,replaces the backbone network in YOLOv5s with a lighter ShuffleNetV2 network,substituted the activation function ReLU in ShuffleNetV2 with H-Swish,and incorporated the CBAM attention mechanism to enhance the model′s focus on key features.Compared with the YOLOv5s model,the average mAP value of this model has increased by 1.5%,while the detection speed has also been shortened by 17.3%.This model can more accurately identify and locate defects,not only improving product quality and production efficiency,but also reducing the workload and error rate of manual inspection.

关 键 词：YOLOv5s 轻量化 自适应融合 轮胎缺陷 目标检测 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]