Table 3.

Classification of macro-expression datasets according to their number of subjects.

Number of Subjects	Macro-Expression Datasets
≤50	TAVER, RAVDESS, BAUM-1, OPEN-EmoRec-II, BP4D-Spontaneous, DISFA, RECOLA, CCDb, MAHNOB Laughter, DEAP, SEMAINE, MAHNOB-HCI, UNBC-McMaster, CAM3D, B3D(AC), MMI-V, AVLC, AvID, AVIC, VAM-faces, ENTERFACE, MMI, MIT, EmoTV, UA-UIUC, 4D CCDb, FreeTalk, IEMOCAP, SAL, iSAFE, ISED
∈[50, 100]	GFT, SEWA, BioVid Emo, MAHNOB Mimicry, AVEC’14, PICS-Stirling ESRC 3D Face Database, Belfast induced (Set2 and Set3), Hi4D-ADSIP, DD, RU-FACS, AAI, Smile dataset
∈[100, 250]	EB+, 4DFAB, AFEW-VA, BP4D+ (MMSE), Vinereactor, CHEAVD, AM-FED, Belfast induced (Set1), USTC-NVIE, CK+
∈[250, 500]	SFEW, Aff-Wild2, AM-FED+, BAUM-2, AVEC’13 AViD-Corpus, DynEmo, AFEW, UT-Dallas
≥500	RAF-DB, AffectNet, Aff-Wild, EmotioNet, FER-Wild, FER-2013, HAPPEI, HUMAINE