Table 6.

Performance of the different deep learning models on the 160-pixel sub-database testing set. The best-achieved results are bold. For the ensemble learning models, WA stands for weighted averaging; UA stands for unweighted averaging, and MV stands for majority voting, and the 3 and 5 at the end of the ensemble models refer to top 3 or 5 base models. All metrics are measured in % unit.

Model	Accuracy	AUC	Precision	Recall	Specificity	F1-Score
MobileNet	98.00	97.72	98.75	96.28	99.17	97.50
MobileNetV2	98.42	98.36	98.03	98.05	98.66	98.04
EfficientNetB0	98.33	98.29	97.83	98.05	98.52	97.94
EfficientNetB1	98.36	98.32	97.85	98.11	98.53	97.98
DenseNet121	98.68	98.58	98.66	98.07	99.09	98.36
DenseNet169	98.57	98.36	99.20	97.25	99.47	98.22
InceptionV3	97.85	97.83	97.02	97.69	97.96	97.36
Xception	97.43	97.34	96.74	96.91	97.78	96.82
Ensemble-WA3	98.94	98.78	99.44	97.94	99.62	98.68
Ensemble-WA5	99.16	99.09	99.19	98.72	99.45	98.96
Ensemble-UA3	99.03	98.93	99.13	98.45	99.42	98.79
Ensemble-UA5	99.20	99.14	99.23	98.80	99.48	99.01
Ensemble-MV3	98.97	98.88	99.06	98.40	99.36	98.73
Ensemble-MV5	99.13	99.07	99.10	98.76	99.39	98.93