基于改进YOLOv5的电子元件表面缺陷检测算法
|
曾耀,高法钦
|
Surface defect detection algorithm of electronic components based on improved YOLOv5
|
Yao ZENG,Fa-qin GAO
|
|
表 2 不同算法在自制数据集上的缺陷检测实验对比结果 |
Tab.2 Experimental comparison results of different algorithms for detection on self-made dataset |
|
模型 | 权重大小/106 | Np /106 | GFLOPs | mAP@0.5 | RF/(帧·s−1) | YOLOv5s | 13.70 | 7.02 | 15.9 | 0.809 | 25 | YOLOv5m | 68.90 | 34.00 | 50.0 | 0.824 | 14 | YOLOv5l | 147.00 | 73.20 | 111.4 | 0.833 | 9 | YOLOv5x | 269.00 | 134.20 | 209.8 | 0.846 | 5 | YOLOv5s-Mobilenet-small | 7.17 | 3.25 | 6.0 | 0.857 | 32 | YOLOv5s-Shufflenet | 1.97 | 3.79 | 7.9 | 0.887 | 34 | GCB-YOLOv5 | 11.60 | 5.90 | 14.0 | 0.930 | 30 |
|
|
|