Development of Emotion Recognition System Using Machine Learning

Law, Soon Kee (2017) Development of Emotion Recognition System Using Machine Learning. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Law Soon Kee .pdf Restricted to Registered users only Download (2MB)

Abstract

The aim of this final year project is to construct an emotion recognition system using the machine learning algorithm. In this project, various algorithms are studied and research in order to select the suitable algorithms to be used in our system. There are several problem to counter in the project such as the difference in background, skin colour, hair style that will increase the difficulty in classify the emotions. In the project, the face detection and extraction stage will be done by using the Haar features based cascade object detector. After that, the mouth and eyes will be detected and extracted again and use as feature. Then, the feature extraction algorithm, local binary pattern (LBP) will be applied on the extracted mouth and eyes images. The LBP image of mouth and eyes will be used to train the classifier. The classifier will be constructed and trained by using the Error Correcting Output Codes (ECOC)-Support Vector Machine (SVM) model that uses the one-against-one coding design. In the project, all images used are image with frontal human face only. The result of face, mouth and eyes are shown. While at the same time, the overall accuracy of the classification is 76.43%.