A Convolutional Neural Network Face Recognition Method Based on BiLSTM and Attention Mechanism

Table 5

Face recognition rates of different algorithms.

Different algorithms	CASIA-FaceV5	LFW	MTFL	CNBC	ORL
ResNet + normalization + centerloss [10]	0.9535	0.9814	—	0.9884	—
MobileNet [11]	—	0.9553	—	—	—
Improved PCA [21]	0.931	0.9446	0.9362	0.9467	0.937
BP [22]	0.8126	0.823	0.8165	0.8217	0.805
PCA [22]	0.8954	0.8789	0.884	0.879	0.854
Gabor + PCA [23]	0.903	0.9186	0.914	0.9231	0.8956
CNN [24]	0.9548	0.959	0.9497	0.9552	0.944
PCANet_dense [25]	—	—	—	—	0.9563
KPCA + SVM [26]	—	—	—	—	0.9516
LBP + PCA + ELM [27]	—	—	—	—	0.91
HOG + LBP [28]	—	—	—	—	0.89
AB-FR	0.9935	0.9646	0.9704	0.9719	0.9679