Real-time emotion detection

• Main Author: Koh, Melvyn Nguan Theng
• Other Authors: Althea Liang Qianhui
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/76147
• Institution: Nanyang Technological University

Deep learning dominates the field of computer vision in recent years and in every few weeks a new deep learning technology takes over the other. Herein, convolutional neural network (CNN) is applied in this project. Detecting facial expressions have been a very fast-growing topic in the field of computer vision as facial expressions are seen as a significant role in human communication and behavioural analysis. Ever since Paul Ekman devised the Facial Action Coding System (FACS) to detect a human facial feature and model the facial behaviours, many scientists are inspired to conduct psychological research on detecting real emotions of a person. Therefore, this has in turn inspired computer scientists to conduct tremendous active research in this field – finding the most accurate and fast models to detecting the true emotion of a person with a camera. This involves using Extended Cohn-Kanade (CK+) and FER2013 datasets. This project aims to build a Real-Time Emotion Detection application that detects seven emotions namely – Anger, Disgust, Fear, Happy, Sad, Surprise and Neutral. The software application is written in Python programming language with OpenCV for processing images and videos. CNN-based approach is done with Google’s Tensorflow machine-learning library to construct the trained model. Lastly, Keras is used as the high-level neural networks API (application programming interface) that runs on top of Tensorflow. The model is trained and evaluated on the FER2013 and CK+ datasets.