Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/63877
Title: Body parts parsing for people in occlusion
Authors: Choon, Hao Wei
Keywords: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Control engineering
Issue Date: 2015
Abstract: Computer vision tools are readily available in our daily life. It conducts series of tasks in the human living environment. This includes the surveillance purposes in the public areas like airport and shopping centres. Due to the complexity danger in the event of terrorism, precise detection of human is necessary for recognizing incoming danger and preventive measures can be imposed. This project aims to develop a system that not only is able to detect human, describing them by breaking them down into different body parts, but also highlight the body parts that have been occluded behind an object, by not showing on the detection of the testing image. With the implementation of HOG, features from the testing image were extracted. Thereafter, by pictorial structure approach, segmentation of the body parts can be achieved. Each individual limbs, head and torso can be detected. Furthermore, a classifier – Supporting Vector Machine (SVM) was used to classify the testing image. The experiments were tested based on 3 factors – namely accuracy, robustness and computational time. With these 3 factors in mind, the implementation of HOG with pictorial structure approach has quite a moderate result. While integrating with a classifier (SVM) for parsing the occlusion body parts, the result improved close to 30% in performance.
URI: http://hdl.handle.net/10356/63877
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:EEE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
HW FYP Report.pdf
  Restricted Access
Final Year Project (FYP)10.25 MBAdobe PDFView/Open

Page view(s) 20

92
checked on Sep 30, 2020

Download(s) 20

9
checked on Sep 30, 2020

Google ScholarTM

Check

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