Emotion recognition using image processing
Emotion recognition using image processing
Date
2020-05
item.page.datecreated
Authors
Frimpong, Queen
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Emotion recognition is an active field of research that has seen a lot of interest over the past
decade. Historically, people’s emotions were analysed and determined through human
observation and psychological counselling and later evolved to electrophysiological and
largely intrusive methods such as Electroencephalography (EEG) because of how complex
of a task it is and how extensive its application could be. Currently, with the entrance of
machine learning and computer vision-related technologies, computers and robots can now
be trained to learn and predict the emotions of human beings either in real-time or with their
static facial images. In this research, emotion recognition is explored with respect to its three
widely recognised stages: face detection, feature extraction and emotion recognition. At each
of these stages, different image processing methods and learning techniques are explored
and tested. A Convolutional Neural Network was trained and tested and recorded an
accuracy of 50.7%. A Support Vector Machine was also trained and tested and recorded an
accuracy of 81.5%. Both classifiers were trained on 7 emotion categories. The results show
that it is possible for computers to predict the emotions of humans using image processing
techniques and deep learning models.
; deep learning; ; image
processing;
Description
Undergraduate thesis submitted to the Department of Computer Science, Ashesi University, in partial fulfillment of Bachelor of Science degree in Computer Science, May 2020
item.page.type
Undergraduate thesis
item.page.format
Keywords
emotion recognition , machine learning , Convotional Neural Network , Support Vector Machine