Table 4. Comparison of multiple different models.

Method	Precision	Recall	Macro-F1
VGG	0.7249	0.7623	0.7642
Inception V3	0.7564	0.7591	0.7845
ResNet	0.7734	0.7611	0.7709
Bi-LSTM	0.7843	0.7882	0.7911
Bi-LSTM-SA	0.8047	0.8014	0.8025