Table 16 Compare the detection performance of different backbone architectures.
From: PEYOLO a perception efficient network for multiscale surface defects detection
Methods | P↑ | R↑ | mAP50↑ | Params↓ | FPS↑ |
|---|---|---|---|---|---|
Baseline with PEYOLO | 0.733 | 0.687 | 75.4 | 3.43 | 194.9 |
Baseline with EfficientViT | 0.626 | 0.701 | 74.2 | 4.15 | 55.3 |
Baseline with Fasternet | 0.681 | 0.727 | 74.9 | 4.32 | 161.5 |
Baseline with LSKnet | 0.711 | 0.639 | 71.7 | 6.04 | 68.4 |
Baseline with Mobilenetv4 | 0.634 | 0.711 | 72.8 | 5.84 | 154.7 |
Baseline with Startnet | 0.675 | 0.706 | 73.5 | 2.36 | 185.2 |
