Please use this identifier to cite or link to this item: http://hdl.handle.net/10266/3454
Title: Age-Group prediction of facial images using different classifiers
Authors: Jain, Madhur
Supervisor: Mishra, Ashutosh
Keywords: age-group prediction;;Viola-Jones algorithm;Support Vector machine;K-Nearest Neighbors;CSE;computer science;software engineering
Issue Date: 30-Jul-2015
Abstract: A human face provides a lot of information which allows another person to identify their characteristics such as age, gender, etc. So the challenge is to develop an age-group prediction system by using the machine learning method. The task of estimating the human‟s age-group from their frontal facial images is very captivating, but also the challenging one due to the personalized and non-linear pattern of ageing which differs from one person to another. This work examines the problem of predicting the age-group of human on the basis of presenting a facial image with the improved accuracy of estimation. The aim of this study is to build up a framework and subsequently an algorithm that helps in estimating the age-group with the reasonable accuracy of the facial images. In this work, a method is presented for the age-group prediction in which age-group is predicted by detecting the face or face landmarks using the Viola-Jones algorithm. After detecting the face, features including geometric features, wrinkles features are extracted and then these extracted features are used to train a classifier using Support Vector Machine (SVM) or K-Nearest Neighbors (K-NN). Finally, SVM or K-NN is used to categorize the age into one of the three different groups such as child, adult and old for the test data. The system used self-build database for the age-group classification. Finally, identification rate achieved using k-NN model produces better results than using SVM model as specified in experimental results.
Description: ME-Software Engineering-Thesis
URI: http://hdl.handle.net/10266/3454
Appears in Collections:Masters Theses@CSED

Files in This Item:
File Description SizeFormat 
3454.pdf1.94 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.