Academic Emotion Classification Using FER: A Systematic Review

Table 5

Summary of commonly used FER datasets.

Dataset	Type of emotions	Data configuration
FER-2013	6 basic facial emotions+1 neutral emotion	(i) Roughly 30,000 grayscale images; (ii) 48×48 scaled facial images with various facial expressions
JAFFE	6 basic facial emotions +1 neutral emotion	(i) 213 images of various facial expressions; (ii) Ten different Japanese females; (iii) Image resolution of
CK+	Happiness, surprise, disgust, sadness, anger, fear, and contempt	(i) 593 video sequences; (ii) 123 different participants with different genders and heritage aged between 18 to 50 years old; (iii) Captured at 30 FPS with a resolution of either or
KDEF	6 basic facial emotions+1 neutral emotion	(i) 4900 photos of human facial emotions from 70 individuals; (ii) Captured from five different angles; (iii) 35 males and 35 females aged between 20 to 30 years old; (iv) Image resolution of; (v) Subjects without beards, eyeglasses, earrings or moustaches, and no noticeable make-up
DISFA	Intensity of 12 AUs coded	(i) Stereo videos of 12 females and 15 males of various ethnic groups; (ii) Image resolution of; (iii) 66 facial landmark points
DISFA+	5-level intensity of twelve FACS	(i) Extension of the DISFA database; (ii) Manually labeled frame-based annotations of 5-level intensity for 12 FACS facial actions