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

Table 1

Facial expression video datasets and emotion categories.

Video dataset	# Videos	Addressed emotions
MMI [21]	2900	Sadness, surprise, fear, happiness, anger, and disgust
CK+ [22]	593	Neutral, sadness, surprise, happiness, fear, anger, contempt, and disgust
DEAP [23]	880	Valence, arousal, and dominance
Belfast [24]	1400	Disgust, fear, amusement, frustration, surprise, anger, and sadness
DISFA [25]	130000	Action unit intensities
RECOLA [26]	3.8 hrs	Valence and arousal
ISED [27]	328	Sadness, surprise, happiness, and disgust
Aff-Wild [ [28]	428	Valence and arousal
BP4D [29]	298	Sadness, surprise, fear, anger, embarrassment, physical pain, happiness, and disgust
RAVDESS [30]	7356	Neutral, calm, happiness, sadness, anger, fear, surprise, and disgust
DEFE [31]	164	Neutral, happiness, anger, valence, arousal, and dominance