Point Based Emotion Classification Using SVM

Abstract

The detection of emotions is a hot topic in the area of computer vision. Emotions are based on subtle changes in the face that are intuitively detected and interpreted by humans. Detecting these subtle changes, based on mathematical models, is a great challenge in the area of computer vision.

In this thesis a new method is proposed to achieve state-of-the-art emotion detection performance. This method is based on facial feature points to monitor subtle changes in the face. Therefore the change in distance between the feature points are extracted. This data is used by a cascade of a multi-class support vector machine, trained to detect all the emotions, and a support vector machine, trained on specific emotions, for binary classification.

This method is implemented in a real-time emotion detection application, with a processing time of less than 30 ms, to show the ability of detecting emotions in real-time.

The Extended Cohn-Kanade (CK+) dataset is used to evaluate the proposed method. The evaluation on this dataset shows that the proposed method has an average accuracy of 90% for detecting 7 basic emotions with outliers for the emotions 'contempt' and 'surprise'. Comparing these results to state-of-the-art algorithms indicates that the proposed method has state-of-the-art accuracies and even outperforms some state-of-the-art algorithms regarding the detection and classification speed of emotions.