Table 8 Performance comparison of each method on GC10-DET datasets.
From: PEYOLO a perception efficient network for multiscale surface defects detection
Method | P↑ | R↑ | mAP50↑ | Params↓ | F1↑ | GFLOPS↓ |
|---|---|---|---|---|---|---|
YOLOv3-tiny | 0.505 | 0.571 | 53.2 | 12.17 | 0.51 | 19.1 |
YOLOv5n | 0.548 | 0.471 | 51.4 | 2.65 | 0.49 | 7.8 |
YOLOv6n | 0.543 | 0.513 | 51.2 | 4.50 | 0.51 | 13.1 |
YOLOv8n | 0.513 | 0.527 | 53.2 | 3.15 | 0.52 | 5.8 |
YOLOv8-ghost | 0.602 | 0.582 | 61.4 | 1.86 | 0.59 | 8.9 |
YOLOv8-world | 0.587 | 0.548 | 57.4 | 4.20 | 0.57 | 39.6 |
YOLOv8-worldv2 | 0.504 | 0.599 | 55.9 | 3.69 | 0.54 | 19.5 |
YOLOv9c | 0.512 | 0.496 | 51.8 | 25.59 | 0.50 | 104 |
YOLOv10n | 0.660 | 0.514 | 56.7 | 2.77 | 0.54 | 8.7 |
PEYOLO | 0.643 | 0.592 | 62.3 | 3.09 | 0.61 | 7.5 |
