Table 14 Compare the detection performance of different feature-fusion network architectures.
From: PEYOLO a perception efficient network for multiscale surface defects detection
Methods | P↑ | R↑ | mAP50↑ | Params↓ | FPS↑ |
|---|---|---|---|---|---|
Baseline with DcPAN | 0.664 | 0.722 | 75.8 | 3.06 | 169.2 |
Baseline with BiFPN | 0.615 | 0.670 | 72.7 | 2.10 | 160.0 |
Baseline with MAFPN | 0.671 | 0.705 | 73.7 | 2.98 | 178.2 |
Baseline with BiFPN-SDI | 0.651 | 0.687 | 73.0 | 2.14 | 130.6 |
Baseline with PAN | 0.668 | 0.710 | 74.8 | 3.15 | 198.1 |
