Table 3.
Comparison of performance scores of YOLOv5 models with different layers and scales.
Method | Precision (%) | Recall (%) | mAP0.5 (%) | mAP0.5:0.95 (%) | Model Size (Byte) | Training Time (min) |
---|---|---|---|---|---|---|
YOLOv5x | 98.3 | 95.2 | 98.0 | 72.4 | 86.2 M | 328.92 |
YOLOv5l | 98.0 | 96.0 | 98.4 | 72.3 | 46.1 M | 199.51 |
YOLOv5m | 97.6 | 96.5 | 98.4 | 71.0 | 20.9 M | 141.41 |
YOLOv5s | 96.0 | 96.2 | 97.8 | 69.6 | 7.00 M | 76.63 |
YOLOv5n + CA(7) | 95.6 | 95.6 | 97.8 | 65.8 | 1.81 M | 52.20 |
YOLOv5n + CA(7)-reduce | 95.5 | 93.4 | 97.6 | 65.1 | 1.72M | 50.30 |