基于全局信息感知的轻量级螺纹钢表面缺陷检测算法

肖剑,杨小苑,何昕泽,陈林,胡欣

Lightweight rebar surface defect detection algorithm based on global information perception

Jian XIAO,Xiaoyuan YANG,Xinze HE,Lin CHEN,Xin HU

表 6 先进主流模型对比实验

Tab.6 Comparative experiments of advanced mainstream models

模型 P/% R/% mAP@0.5/% mAP@0.5-0.95/% Np/M FLOPs/G FPS/(帧·s−1) Faster R-CNN 82.1 72.9 82.7 44.6 137.1 370.2 6 SSD 80.6 80.7 85.0 49.3 41.1 145.3 41 YOLOv3-Tiny 80.8 79.9 86.3 50.9 12.1 18.9 78 YOLOv4-Tiny 84.3 81.0 88.4 52.1 5.9 16.1 113 YOLOv5n 85.5 81.2 89.4 52.5 2.4 7.1 107 YOLOv6n 81.6 80.6 87.2 51.3 4.2 11.2 130 YOLOv7-Tiny 85.4 81.2 89.3 54.7 6.1 13.1 109 YOLOX-Tiny 86.2 82.5 90.3 56.1 5.1 6.5 122 YOLOv8n 88.3 83.1 90.2 55.9 3.0 8.1 133 YOLOv9n 85.5 83.3 90.7 56.5 2.3 8.4 135 YOLOv10n 85.6 82.7 90.5 55.9 2.7 8.2 131 YOLOv11n 86.1 83.4 90.9 55.8 2.6 6.3 136 YOLOv8-VSC 88.6 88.8 92.7 56.3 2.0 6.0 152 LTSCD-YOLO 89.2 87.8 91.9 56.5 2.4 9.8 144 S-YOLO 89.8 87.9 93.2 57.9 2.6 8.4 127 本研究模型 90.2 89.0 94.5 58.8 1.4 5.3 156