Table 1 Training and validation results of the models.
Model | Stage | Precision% | Recall% | F1-score% | mAP% | Accuracy% |
|---|---|---|---|---|---|---|
YOLOv4 | Training | 87.08 | 67.55 | 76.15 | 65.06 | 82.79 |
Adaptive light-weight YOLOv4 | Training | 84.79 | 82.61 | 83.68 | 80.73 | 81.44 |
SSD | Training | 83.70 | 75.73 | 79.52 | 71.10 | 75.97 |
ASODE model | Training | 85.24 | 83.81 | 84.52 | 78.58 | 84.66 |
YOLOv4 | Validation | 84.88 | 83.80 | 84.16 | 86.49 | 93.91 |
Adaptive light-weight YOLOv4 | Validation | 89.25 | 80.76 | 85.15 | 76.36 | 85.80 |
SSD | Validation | 82.39 | 76.29 | 80.76 | 74.76 | 83.71 |
ASODE model | Validation | 90.29 | 87.96 | 89.38 | 85.31 | 88.29 |
