Table 7.

Performance of the state-of-the-art CNN adapted via cross-domain TL on test dataset.

Classification model	F-score ± S.E	Precision	Recall
Mit-VGG	0.67 ± 0.027	0.61	0.74
Mit-ResNet	0.71 ± 0.026	0.66	0.77
Mit-DenseNet	0.72 ± 0.026	0.65	0.82
Mit-ResNext	0.72 ± 0.026	0.68	0.75
Mit-SeNet	0.66 ± 0.027	0.64	0.69
Mit-PolyNet	0.65 ± 0.027	0.58	0.74