Emotion recognition using anatomical information in facial expressions

This paper shows a work done under Affective Computing umbrella and in the field of emotion recognition. The paper explores the anatomy of a human face and builds the classification model based on it. The anatomical information of face is used to locate several points on the face and to extract the features. The features are in form of distance vectors which can be of specific person or group of persons. Canny edge detection algorithm is used to refine the extracted features and make them suitable for classifications. For classification two approaches (one vs. rest and one vs. one) of SVM (Support Vector Machine) classification for multiple classes (happy, sad, disgust, anger and surprise) have been incorporated. The paper shows the results and analysis of both these methods.