Table 8 Per-class detection and classification results of the proposed model.
From: An intelligent YOLO and CNN-BiGRU framework for road infrastructure based anomaly assessment
Class | Precision | Recall | F1-score | AUROC |
|---|---|---|---|---|
Surface cracks | 0.962 | 0.948 | 0.955 | 0.987 |
Faded/missing markings | 0.953 | 0.939 | 0.946 | 0.983 |
Potholes | 0.936 | 0.918 | 0.927 | 0.978 |
Severe structural damage | 0.901 | 0.854 | 0.877 | 0.961 |
Snow/Ice covered roads | 0.882 | 0.826 | 0.853 | 0.948 |
Macro average | 0.927 | 0.897 | 0.912 | 0.971 |
Weighted average | 0.948 | 0.934 | 0.941 | 0.981 |
