一种轻量级道路缺陷检测算法OA

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.