Face detection on Our dataset of real surgeries
89.51PrecisionIter-Score
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Iter-ScoreBase Detector=P-D-DETR2026.02 | 89.51 | 92.12 | 88.86 | 61.97 | 87.12 | |
| Our approach (P-D-DETR)Base Detector=P-D-DETR, FPS=8.632026.02 | 87.58 | 97.15 | 88.22 | 84.44 | 95.33 | |
| Our approach (DEIM)Base Detector=DEIM [15], FPS=52.092026.02 | 86.21 | 97.55 | 88.3 | 86.97 | 96.06 | |
| P-D-DETRTraining Dataset=CrowdHuman2026.02 | 84.38 | 93.13 | 88.67 | 70.21 | 88.8 | |
| RetinaFaceTraining Dataset=WIDER FACE2026.02 | 66.3 | 39.25 | 31.99 | 8.91 | 22.87 | |
| Head-YOLOv8Training Dataset=CrowdHuman2026.02 | 33.97 | 44.82 | 18.62 | 12.37 | 29.56 |