Table 9 Performance evaluation and comparison of various models on the NTHU-DDD dataset.
Model | Precision (in %) | Recall (in %) | F1 Score (in %) | Accuracy (in %) |
|---|---|---|---|---|
ViT Transformer | 99 | 100 | 100 | 99.52 \((\pm 0.09\)) |
Swin Transformer | 99 | 99 | 99 | 98.76 \((\pm 0.11\)) |
VGG19 + Attention | 99.44 | 98.52 | 98.98 | 99.07 \((\pm 0.12\)) |
VGG19 | 98 | 98 | 98 | 98.66 \((\pm 0.13\)) |
DenseNet169 | 98.52 | 98.44 | 98.48 | 98.60 \((\pm 0.12\)) |
ResNet50V2 | 97.41 | 99.48 | 98.43 | 98.55 \((\pm 0.13\)) |
InceptionResNetV2 | 98.74 | 94.70 | 96.68 | 97.02 \((\pm 0.14\)) |
InceptionV3 | 98.71 | 97.74 | 98.22 | 98.38 \((\pm 0.13\)) |
MobileNet | 90.20 | 89.52 | 89.86 | 91.15 \((\pm 0.12\)) |
