Table 5.

The result of accuracy, precision, recall, F1-score on average and the number of layers for all experiments

Network	Acc. (%)	Prec. (%)	Rec. (%)	F1. (%)	# of layers
W-Net	97	97	97	97	3
W-Net-SVM	95	97	95	96	3
AlexNet	84	94	84	85	8
VGGNet	44	67	44	42	16
ResNet50	51	60	51	43	50
ResNet18	79	81	78	77	18
RNN	83	86	85	85	–

The parts in bold mean our model