汽车工程学报2026，Vol.16Issue(3)：395-405,11.DOI:10.3969/j.issn.2095-1469.2026.03.06
一种轻量级道路缺陷检测算法
A Lightweight Algorithm for Road Defect Detection
摘要
Abstract
To address the issues of insufficient accuracy,large number of parameters,and the difficulty of deploying existing algorithms on mobile devices,a lightweight road defect detection algorithm named RDD-YOLOv8 is proposed.Firstly,the RepHGNetV2 was introduced into the backbone network to maintain detection accuracy while achieving a lightweight model design.Secondly,the SimAM attention mechanism was added to enhance the overall performance of the model.Finally,the convolution module in the neck network of YOLOv8n was replaced with GSConv,and the VoVGSCSP module was constructed accordingly.The neck network was then redesigned as a Slim-neck structure composed of GSConv and VoVGSCSP,further improving the lightweight design of the model.Through experiments on the RDD-2022 dataset,RDD-YOLOv8 shows an increase of 0.7 percentage points in both mAP0.5 and mAP0.5:0.95 compared to the original YOLOv8n algorithm.Meanwhile,Flops decrease by 24.7％and the number of parameters decreases by 29.2％.These results show that the proposed algorithm significantly enhances the real-time detection performance.关键词
道路缺陷检测/轻量级算法/无参数注意力机制/Slimneck模块/主干网络Key words
road defect detection/lightweight algorithm/SimAM/Slimneck/RepHGNetV2分类
交通工程引用本文复制引用
路成龙,庆光蔚,孙勇,倪大进..一种轻量级道路缺陷检测算法[J].汽车工程学报,2026,16(3):395-405,11.基金项目
国家市场监督管理总局科技计划项目(2024MK168) （2024MK168）
江苏省市场监督管理局科技计划项目(KJ2025049) （KJ2025049）
