Laughter emotion recognition using gestures

Date of Publication

2014

Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Computer Science

Subject Categories

Programming Languages and Compilers

College

College of Computer Studies

Department/Unit

Computer Science

Thesis Adviser

Cu, Jocelyn

Defense Panel Chair

Merlin Teodosia Suarez

Defense Panel Member

Jocelyn Cu
Rafael Cabredo

Abstract/Summary

Laughter is a form of communicative response by humans, and there are many forms of laughter as well as gestures indicating what kind of laughter however there are few studies concentrating on it. With the use of the body-tracking technology in Microsoft Kinect, it is possible to detect the movements of the subject while they are laughing.

These points are then taken and computed into features such as head tilt, shift in seat, leaning, shoulder shift center, shoulder shift right and shoulder shift left. Then, this data is fed into Weka using three algorithms namely C.45, kNN and SVM with Polykernel, Puk and RBF kernel and tested using 10-fold cross validation with combined, individual, adjusted and male/female datasets. SVM was classified with its own adjusted dataset.

The highest correctly classified instances between kNN and C4.5 was at 95.5% with kappa statistic of 0.9435 and RMSE of 0.1348, from Natutuwa adjusted instances dataset. With SVM, it is SVM with Puk kernel that performed the best, with 77.0% correctly classified instances, kappa statistic of 0.5405 and an RMSE 0.4793 from the comparison of nasasabik.

Abstract Format

html

Language

English

Format

Electronic

Accession Number

CDTU019261

Shelf Location

Archives, The Learning Commons, 12F, Henry Sy Sr. Hall

Physical Description

leaves ; 4 3/4 in.

This document is currently not available here.

Share

COinS