Table 1 The performance of emotion recognition was evaluated in each modality (EEG, Visual, and Audio data).

Database	Primary Modalities	Language	Subjects	Elicitation Method	Types
MAHNOB-HCI⁶	EEG, Face, Audio	—	27 subjects	Videos/Pictures	S, I
SEED-IV¹¹	EEG and EM	—	15 subjects	Videos	S, I
DREAMER¹⁰	EEG & ECG	—	23 subjects	Movies	S, I
MPED⁹	EEG, GSR, RR, ECG	—	23 subjects	Videos	S, I
ASCERTAIN⁸	EEG, ECG and GSR	—	58 subjects	Videos	S, I
AMIGOS⁷	EEG, GSR and ECG	—	40 subjects	Movies	S, I
DEAP¹²	EEG, PS, Face	—	32 subjects	Music videos	S, I
IEMOCAP¹⁵	Face, Speech, Head	English	10 professional actors	Conversations	S, I
SEMAINE¹⁶	Face, Speech	English	150 subjects	Conversations	S, I
NNIME¹⁹	Audio, Video, ECG	Chinese	44 subjects	Conversations	P, N
RAVDESS²⁰	Audio, Video	English	24 professional actors	Speech, Song	P, I
BAUM-1²¹	Face, Speech	Turkish	31 subjects	Images/Videos	S, I
SAVEE²⁶	Face, Speech	English	4 subjects	Videos/Texts/Pictures	S, I
K-EmoCon¹⁴	Face, Speech, 1ch EEG	Korean	32 subjects	Conversations (Debate)	S, N
PEGCONV¹³	EEG, GSR, PPG	English	23 subjects	Conversations	S, N
EAV (ours)	EEG (30ch), Audio, Video	English	42 subjects	Conversations	S, I

Specifically, accuracy and AUC scores for 5 balanced classes were calculated across individual subjects.

Quick links

Search