The comparative study of model-based and appearance Based gait recognition for leave bag behind

Zainol, Norfazilah (2018) The comparative study of model-based and appearance Based gait recognition for leave bag behind. Masters thesis, Universiti Tun Hussein Onn Malaysia.

[img]
Preview
Text
24p NORFAZILAH ZAINOL.pdf

Download (16MB) | Preview
[img] Text (Copyright Declaration)
NORFAZILAH ZAINOL COPYRIGHT DECLARATION.pdf
Restricted to Repository staff only

Download (1MB) | Request a copy
[img] Text (Full Text)
NORFAZILAH ZAINOL WATERMARK.pdf
Restricted to Registered users only

Download (53MB) | Request a copy

Abstract

Nowadays, the increasing number of crimes and violence in the world has become a concern of modem society. This is why the need for criminal recognition using gait used for civilian and forensic analysis applications has evoked considerable interest. The literature accurate the result can be found in gait recognition by leave bag behind detection especially in .the critical area examples airport and shopping mall environment. This is important because the method used capable of identifying the subject based on their gait and can be presented as the most probable subject as a strong evidence for criminal identification. This research limited to leave the bag behind detection on gait recognition. In this research, the analysis perfonned using two methods which are Model-Based approaches and Appearance-Based approaches. The selected methods were implemented in MATLAB R2014a and R Studio and tested with a standard dataset from the Chinese Academy of Science (CASIA) and tested using two classifiers which is Support Vector Machine (SVM) and KNN (K nearest Neighbour) based on accuracy and misclassification rates (MER) metrics. The experiment results show that the accuracy and misclassification rate (MER) of Appearance-based approaches obtained is 93.66% and 6.33% respectively tested on SVM classifier then the accuracy and misclassification rate (MER) of Appearance­based approaches is 97.66% and 2.33% respectively tested on KNN algorithm. Meanwhile, the accuracy and misclassification rate (MER) of Model-based approaches obtained is 97.00% and 3.00% respectively tested on SVM classifier then the accuracy and misclassification rate (MER) of Model-based approaches is 99.00% and 1.00% respectively tested on KNN algorithm. It can be concluded from experiments conducted by Model-based approaches better than Appearance-based approaches because Model-Based approaches higher precision value as well as low misclassification.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics
Q Science > QA Mathematics > QA76 Computer software
Divisions: Faculty of Computer Science and Information Technology > Department of Software Engineering
Depositing User: Mrs. Sabarina Che Mat
Date Deposited: 21 Aug 2022 01:44
Last Modified: 21 Aug 2022 01:44
URI: http://eprints.uthm.edu.my/id/eprint/7551

Actions (login required)

View Item View Item