Table 1 Fissures.
From: Lightweight YOLOv8 for tongue teeth marks and fissures detection based on C2f_DCNv3
Network | mAP | F1 | Precision | Recall |
|---|---|---|---|---|
Fater-RNN | 86.00% | 0.75 | 65.71% | 88.46% |
Efficientdet | 87.56% | 0.85 | 88.43% | 82.31% |
Centernet | 91.26% | 0.83 | 92.12% | 73.85% |
Yolov8 | 86.10% | 0.85 | 89.66% | 80.00% |
Yolov8+CBAM | 87.20% | 0.86 | 88.11% | 81.54% |
Yolov8+ECA | 87.30% | 0.85 | 87.70% | 82.31% |
Yolov8+SE | 89.23% | 0.86 | 88.62% | 83.85% |
Yolov8+C2f_DCNv3 | 87.79% | 0.83 | 86.36% | 80.69% |
Yolov8+C2f_DCNv3+SE | \(\pmb {91.34\%}\) | 0.87 | 88.35% | 84.94% |
