Table 7 Comparison experiment results.
From: Improved YOLOv8n-based bridge crack detection algorithm under complex background conditions
Number | Model | GFlops | Parameters | Recall | Precision | mAP@0.5 | mAP@0.5–90 |
|---|---|---|---|---|---|---|---|
1 | YOLOv5s | 24.04 | 9,122,579 | 76.31% | 84.67% | 81.63% | 62.92% |
2 | YOLOv6s | 44.21 | 16,306,035 | 76.71% | 88.48% | 82.37% | 64.90% |
3 | YOLOv8-odconv | 7.04 | 3,045,615 | 74.70% | 87.57% | 82.44% | 63.65% |
4 | YOLOv8-Gost | 5.14 | 1,719,159 | 77.87% | 83.98% | 81.98% | 63.42% |
5 | YOLOv9c | 103.68 | 25,530,003 | 76.71% | 87.95% | 84.01% | 64.95% |
6 | Ours | 16.89 | 4,608,707 | 77.11% | 88.00% | 84.66% | 66.83% |
