Comparison of Tuplet of Techniques for Facial Emotion Detection

Abstract

Intelligent systems use real-time emotion recognition for interacting with humans in a better way to dynamically adapt their responses based on user emotions. Emotion detection is a combination of facial expression, gestures and speech features. With proper emotion detection using facial features, the mental health of subjects could be monitored and improved, thus preventing unruly actions. Machine learning techniques applied for feature extraction and classification would perform emotion detection based only on facial expression. In this work accuracy of up to 0.9042 was achieved using a combination of Latent Dirichlet Allocation (LDA) + Support Vector Machines (SVM).