Table 1 Comparison of literatures

CK+

Liu et al. [11]

DBDN

96.7

Shan et al. [21]

LBP of weighted patch

SVM

88.4

Happy et al. [23]

LBP of active patch

SVM

94.09

Lopes et al. [28]

CNN

96.76

Zhong et al. [35]

LBP

Multi-task sparse learning

89.89

Shan et al. [45]

Boosting LBP

SVM

92.6

Burkert et al. [47]

CNN

99.6

Ding et al. [49]

FaceNex2ExpNet

98.6

Zeng et al. [50]

LBP, HOG, gray value

DSAE

95.79

Zeng et al. [51]

LTNet

92.45

Chen et al. [52]

HOG-TOP

SVM

89.6

Kumari et al. [53]

LBP, HOG, LDP

K-Nearest Neighbor

97.96

JAFFE

Liu et al. [11]

DBDN

91.8

Happy et al. [23]

LBP of active patch

SVM

92.63

Lopes et al. [28]

CNN

53.57

Shan et al. [45]

Boosting LBP

SVM

81

Kumari et al. [53]

LBP, HOG, LDP

K-Nearest Neighbor

95.31

MMI

Zhong et al. [35]

LBP

Multi-task sparse learning

73.53

Shan et al. [45]

Boosting LBP

SVM

86.9

Burkert et al. [47]

CNN

98.63

Zeng et al. [51]

LTNet

65.61