Table 3 mAP@0.5 and GFLOPS.
From: Lightweight YOLOv8 for tongue teeth marks and fissures detection based on C2f_DCNv3
Network | mAP | GFLOPS |
|---|---|---|
Fater-RNN | 83.76% | 970.21G |
Efficientdet | 86.98% | 26.17G |
Centernet | 91.20% | 70.22G |
Yolov8 | 90.89% | 28.81G |
+CBAM | 90.74% | 28.82G |
+ECA | 91.22% | 28.82G |
+SE | 91.02% | 28.82G |
+C2f_DCNv3 | 90.78% | \(\pmb {21.01G}\) |
+C2f_DCNv3+SE | \(\pmb {92.77\%}\) | \(\pmb {21.01G}\) |
