Table 4. Results of the DUF classification task trained with 10,000 images.

Model	Accuracy (%)	Specificity (%)	Recall (%)	Precision (%)	F1-Score (%)
Proposed model	97.5	95.5	99.6	95.4	97.5
Xception	98.9	98.2	99.6	98.2	98.9
ResNet50	99.4	98.9	100	98.9	99.4