Table 4. Classification models with frozen layers.

Parameters	VGG16		MobileNetV2		ResNet50
Epochs	10	20	10	20	10	20
Execution time (seconds)	4,517	5,251	1,438	3,203	3,101	4,133
Early stopped?	No	Yes, at Epoch 13	No	No	No	Yes, at Epoch 14
Training accuracy (%)	99.8	99.9	98.2	99.99	99.8	99.9
Validation accuracy (%)	97.4	97.6	87.6	98.98	99	99.2
Testing accuracy (%)	94.03	94.1	80.84	95.20	97.3	96.43
Precision	0.95	0.96	0.83	0.95	0.97	0.97
Recall	0.94	0.94	0.8	0.95	0.97	0.96
F1-Score	0.94	0.94	0.8	0.95	0.97	0.96