Table 2.

Comparison of the proposed architecture within the well-known CNN architectures and the baseline [22]

	Accuracy (%)	Recall (%)	Specificity (%)	F1-score (%)	g-mean (%)
AlexNet [24]	66.23	66.33	50.25	67.56	57.73
VGG16 [25]	60.45	62.56	52.22	62.54	57.15
VGG19 [25]	59.03	64.87	59.63	61.39	62.19
GoogleNet [26]	76.55	77.23	65.97	78.74	71.37
CNN-LSTM [7]	81.35	79.05	80.22	79.63	79.63
CNN-BiLSTM [8]	80.87	75.33	81.14	78.12	78.05
DenseNet [22]	84.70	76.20	–	85.30	–
Proposed architecture	87.56	93.81	81.17	87.92	87.56