Table 5.

Assessment of classification models with attention mechanism.

Method	Accuracy (%)	Precision (%)	Recall (%)	F-1 score (%)	Computation time per validation image (ms)
RNN with an attention mechanism	96.10	95.22	95.89	95.55	49
DBN with an attention mechanism	95.26	96.19	96.95	96.57	43
LSTM with an attention mechanism	97.01	96.52	96.61	96.56	35
T-LSTM with an attention mechanism	99.98	99.76	99.89	99.82	20