MULTIMODAL EMOTION RECOGNITION BASED ON FEATURE FUSION FOR ENHANCEMENT OF HUMAN-COMPUTER INTERACTION

Abstract: Emotions are intrinsically part of human mental activity and play a key role in human decision handling, interaction and cognitive processes. Recognizing emotion is an essential step to have complete interaction between human and machine. Emotion Recognition (ER) frameworks is especially significant for a human personal relationship. Feelings are created by some physiological changes. The clear view of this exertion is to find the capability of language and facemask components to convey the inclination precise data for improving the Human-Machine interaction. The methods and frameworks utilized in emotion detection may differ depending on the features reviewed. The combination of features is performed either at the decision level or the previous arrangement. Multimodal approaches by consolidating the method of cooperation brings about upgrading the level and outcomes in a productive ER framework as far as better execution and power. Since both these features compete one another, consolidating them brings about better as far as precision of 94.843%. The proposed framework was tried on ENTERFACE dataset and ongoing video. For Video, Speeded Up Robust Features (SURF) and Gabor highlights are utilized. Keywords-: Emotion recognition; Multimodal Approach; Support Vector Machine (SVM); SURF and Gabor features;