Please use this identifier to cite or link to this item:
|Title:||Automated emotion recognition based on extreme learning machines||Authors:||Tan, Yikai||Keywords:||DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems||Issue Date:||2014||Abstract:||Although the information in still images can already enable a computer to perform emotion recognition, it is only natural for moving images tocontain even more information, empowering the computer to further improve its ability to recognize emotions. Hence, in this project, we will investigate the effectiveness of emotion recognition using visual information from videos, by using dynamic Haar-like filters for feature extractions and Extreme Learning Machine (ELM) as the classifier. The project will be segmented into 3 parts. In the first part we will be looking into Static haar-like features, while in the second part we will be looking into dynamic haar-like features. Both part 1 and 2 contains 3 phases, pre-processing, feature extraction, classification. In first phase, pre-processing, facial normalization based on eye coordinates will be performed on images from the Cohn-Kanade Database. The second phase is feature extraction, for still images, static haar-like filters will be used, to extract static haar-like features from the most expressive normalized images of each subject, while for moving images, dynamic haar-like filters will be used to extract dynamic haar-like features from the normalized video sequence of each subject The third phase is classification, where training and testing will be done on extracted features using Extreme Learning Machine with kernel to evaluate accuracy of emotion classifications. In part 3, accuracy results of both static and dynamic haar-like features will be compared to test effectiveness of using dynamic haar-like features. Finally, further integration will be done on 2 different classifiers in relation to dynamic haar-like features, namely Sparse Representation Classifier (SRC) and Extreme Learning Machine (ELM).||URI:||http://hdl.handle.net/10356/60112||Rights:||Nanyang Technological University||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Student Reports (FYP/IA/PA/PI)|
Updated on Dec 1, 2020
Updated on Dec 1, 2020
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.