Indonesian Lip-Reading Detection and Recognition Based on Lip Shape Using Face Mesh and Long-Term Recurrent Convolutional Network

Table 5

Results of the performance evaluation (precision, recall, and F1 Score) on the phrase dataset.

Phrase	Conv-LSTM			LRCN-2Conv			LRCN-3Conv
	Precision (%)	Recall (%)	F1score (%)	Precision (%)	Recall (%)	F1score (%)	Precision (%)	Recall (%)	F1score (%)
Terima kasih	98	100	99	100	97	99	100	97	99
Saya minta tolong	86	80	83	95	88	91	93	93	93
Saya minta maaf	88	90	89	97	97	97	97	97	97
Minta tolong	90	93	91	89	97	93	93	95	94
Accuracy			91			95			96
Macro avg	91	91	91	95	95	95	96	96	96
Weighted avg	91	91	91	95	95	95	96	96	96