Ensemble Convolution Neural Network for Robust Video Emotion Recognition Using Deep Semantics

Table 14

Accuracy of maximum ensemble of different techniques.

Method	Accuracy
	Angry	Disgust	Fear	Happy	Sad	Surprise	Neutral	Overall
CNNHOG-KLT	92.93	90.91	89.00	92.86	91.20	91.09	91.75	91.39
CNNHaar-SVM	89.24	89.31	90.91	91.12	91.08	91.92	93.72	91.04
CNNPATCH	90.09	88.35	93.68	90.19	91.00	90.09	91.84	90.75
CNNENSEMBLE	92.93	90.91	90.91	92.86	91.20	91.92	93.75	92.07

Bold values show that the accuracy value of all emotions are more accurate in the CNN-ENSEMBLE model as compared with other models.