基于改进YOLOv8s的钢材表面缺陷检测算法

梁礼明,龙鹏威,金家新,李仁杰,曾璐

Steel surface defect detection algorithm based on improved YOLOv8s

Liming LIANG,Pengwei LONG,Jiaxin JIN,Renjie LI,Lu ZENG

表 5 NEU-DET和Severstal数据集消融实验结果

Tab.5 Ablation test results on NEU-DET and Severstal datasets

数据集 模型方法 mAP/% Params/106 FLOPs/109 FPS/帧 P/% R/% NEU-DET YOLOv8s 72.8 11.1 28.8 109 72.8 70.9 YOLOv8s+M1 77.2 10.3 26.9 129 77.1 70.4 YOLOv8s+M2 75.8 12.3 28.7 104 76.7 70.2 YOLOv8s+M3 74.0 6.4 16.5 217 71.5 76.0 YOLOv8s+M1+M2+M3 79.2 7.2 16.2 146 77.4 72.1 Severstal YOLOv8s 69.9 11.1 28.8 103 71.1 68.4 YOLOv8s+M1 74.2 10.3 26.9 112 70.0 69.7 YOLOv8s+M2 71.1 12.3 28.7 88 68.8 70.9 YOLOv8s+M3 72.1 6.4 16.5 196 65.1 70.3 YOLOv8s+M1+M2+M3 76.9 7.2 16.2 121 77.6 70.4