Table 3 Detection performance on road condition dataset.
From: An intelligent YOLO and CNN-BiGRU framework for road infrastructure based anomaly assessment
Model | mAP@0.5 (%) | mAP@0.5:0.95 (%) | Avg. AUROC | F1-score |
|---|---|---|---|---|
Mask R-CNN | 89.7 | 68.9 | 0.931 | 0.881 |
Detectron2 | 91.4 | 70.2 | 0.944 | 0.896 |
YOLOv7 | 91.3 | 70.4 | 0.948 | 0.901 |
YOLOv8 | 93.2 | 72.9 | 0.958 | 0.916 |
YOLOv9 | 94.1 | 74.7 | 0.963 | 0.927 |
YOLOv10 | 95.3 | 76.8 | 0.967 | 0.933 |
YOLOv11 | 96.1 | 78.1 | 0.972 | 0.941 |
Proposed model | 97.5 | 80.4 | 0.982 | 0.954 |
