Table 6 Comparative experiments on improving networks and other networks on aluminum datasets.
From: Aluminum surface defect detection method based on a lightweight YOLOv4 network
Network model | P | M | Op | Cr | Nc | R | F | S | |
|---|---|---|---|---|---|---|---|---|---|
YOLOv4 | 93.31 | 1 | 1 | 0.99 | 0.97 | 0.95 | 0.95 | 0.85 | 0.75 |
M2-YOLOv4 | 92.25 | 1 | 1 | 1 | 0.99 | 0.95 | 0.94 | 0.83 | 0.67 |
YOLOv5 | 89.33 | 0.98 | 0.98 | 0.84 | 0.98 | 0.86 | 0.97 | 0.69 | 0.82 |
YOLOv7 | 92.25 | 1 | 0.98 | 0.86 | 1 | 0.9 | 0.97 | 0.86 | 0.81 |
Li’s method26 | 87.38 | 1 | 0.98 | 0.98 | 1 | 0.83 | 0.93 | 0.73 | 0.54 |
Hao’s method27 | 90.75 | 1 | 0.97 | 0.97 | 0.92 | 0.92 | 0.91 | 0.89 | 0.68 |
M2-BL-YOLOv4 | 93.5 | 1 | 1 | 1 | 0.99 | 0.96 | 0.95 | 0.85 | 0.73 |
