From: Facial expression recognition using three-stage support vector machines
Datasets | Method | Classification rate | Proposed method | Validation test |
---|---|---|---|---|
JAFFE | Patch-based Gabor [5] | 92.30% | 96.71% | Leave-one-out |
HOG+SVM [5] | 94.30% | |||
LDA + SVM [4] | 95.71% | |||
Boosted LBP + SVM [3] | 79.80% | 98.10% | 10-folds | |
LBP pyramid + SVM [1] | 91.36% | |||
LBP + Gabor filter [1] | 92.38% | |||
LBP + HOG + PCA + SVM [6] | 87.60% | 97.62% | 5-folds | |
CK+ | HOG+SVM [5] | 88.70% | 93.29% | Leave-one-out |
DNN [7] | 93.20% | 93.49% | 5-fold | |
RaFD | HOG+SVM [20] | 98.50% | 99.72% | 10-fold |