Please use this identifier to cite or link to this item:
Title: Emotion recognition - 1
Authors: Muhammad Farhan Othman
Keywords: DRNTU::Engineering::Computer science and engineering
Issue Date: 2017
Abstract: Emotion recognition has become an emphasising research area since it plays an important role in the human computer interaction paradigm. However, emotions can be complex which makes it challenging for machines to perceive various facial expressions. Unlike humans, machines require a much subtle approach in order to classify a facial expression correctly. This project aims to build a real-time Emotion Recognition program which detects four basic emotions: Happy, Anger, Sadness and Surprise. To achieve the aforementioned, the CK+ database will be used for training the emotion and Action Units classifiers. In addition, the Microsoft Kinect V1 Sensor will be used to track the facial features in real-time. Finally, the detected facial features will be classified to obtain the corresponding emotion and action units. To evaluate the performance of the Emotion Recognition program, test subjects are required to watch a series of videos which are aimed to evoke the desired emotion. The results are recorded and will be presented in this report. Also, this report will highlight the performance of the classifiers with the different machine learning algorithms such as Naïve-Bayes, Decision-Tree and Support Vector Machine. It was shown that for emotion classification, the one vs one Multi-Class SVM with the Gaussian kernel is most suited. For Action Units classification, a combination of Multi-label and Multiclass SVM with the Gaussian kernel shows acceptable performance.
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:SCSE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
Muhd_Farhan_Bin_Othman_FYP_Report .pdf
  Restricted Access
Final year Report2.6 MBAdobe PDFView/Open
  Restricted Access
222 BHTMLView/Open

Page view(s) 10

Updated on Nov 25, 2020

Download(s) 10

Updated on Nov 25, 2020

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.