Identification of Facial Gestures and Audio Visual Interactions Using Ensemble Matrix of Multi Classifiers

Human emotions are expressed using the facial expressions, the tone of voice, hands and body gestures .An approach of interaction between the computer-human must be accurate and robust. We use the concept of multi classifier systems to study the human emotions, audio-visual detection system. Automate recognition of emotional state, machines must be taught expressions to understand facial gestures. Here a proposal to identify the person's emotion state such as happy, anger, disgust etc. is stated. Audio visual detection is performed by fusing the results of separate audio and video classifiers on the decision level. We also study about the interactive visualization using Ensemble Matrix. It provides an approach for users to directly interact with the visualization in order to explore and build combination models. The efficiency of the system and approach in a user study is done. Multiple classifiers can be combined on multiple feature set to produce an ensemble classifier with accuracy that will provide a best-reported performance. Keywords—Emotions, Audio-Visual, Ensemble Matrix, SVM classifier, facial gestures