Real-time emotion detection using python

Abstract

The thesis aims to learn more about the machine learning model and develop the model to explore the use of computer vision techniques using Python and Convolutional Neural Networks (CNNs). By developing an CNNs model to learn about human basic emotions through Kaggle's FER2013 dataset, a deep learning model can classify emotions based on facial expressions on real-time video feed and display an accuracy percentage along with emotion labels in performance. Through all the tests and the ability to recognize emotions in real-time, the emotion detector can capture and analyses the emotions through webcam and shows the accuracy rates in each expression that has been made and recorded.

The thesis consisted three parts, which included theoretical sections, practical sections and evaluation. The theoretical will have introduction about Deep Learning and Imagine Processing. The practical part will showcase toward a program that can detect human faces and recognize their facial emotion through cameras in real time with Deep Learning methods for face recognition and Python Programming Language. After the successful test running, the project of Emotion Detection will using webcam to determine user’s facial expression with accuracy rate and emotion labels.