Automatic facial expression recognition on smartphone

In a variety of sectors, automatic facial expression recognition (AFER) has seen increased use in recent years. With the success of face detection for unlocking screens on smartphones. Implementing the facial expression recognition (FER) system that can be used on smartphones will enable the development of more interesting applications, such as games and other useful mobile applications. However, implementing an AFER application on a smartphone is a challenging task, because traditional human emotion algorithms are usually computationally intensive and only can be implemented offline on a computer. Therefore, this paper presents an AFER mobile application, the FER mobile application is a real-time running on the smartphone with the mobile camera. The proposed FER application is to use a Convolutional Neural Networks (CNN) for classification of six basic emotions plus contempt. For the facial expression detection and features are extracted by HAAR Cascade Classifier and the result of the classification will be displayed on the screen immediately. The experiment shows a result of 69.5% of the test accuracy. The experiment is using the Cohn-Kanade (CK+) dataset which include 593 video sequences from 123 different subjects.Bachelor of Engineering (Computer Science