Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/15790
Title: Speeding up boosted cascade of object detection using commodity graphics hardware
Authors: FENG JIMIN
Keywords: GPGPU, object detection, data streaming paradigm, Boosted learning, predication, cascaded detector
Issue Date: 28-May-2009
Source: FENG JIMIN (2009-05-28). Speeding up boosted cascade of object detection using commodity graphics hardware. ScholarBank@NUS Repository.
Abstract: Visual object detection has conceivably prevailing applications in the Internet age for multimedia interactions. This thesis aims to explore the data-parallel architecture of commodity graphic processors that enables fast and promising object detection. Firstly, Boosting is identified as the promising and widely usable approach for object detection. An efficient architecture in streaming paradigm is then designed to map the boosted cascade to GPU as data streaming coprocessor. We take into ccount the hardware characteristics to enable further speedup. Promising results were achieved, as up to 5 times speedup was obtained. By study the performance impacts from experiments on GPUs that are of different generations, we explore the suitable designs for future generation GPU models. Our experiences reveal the underlying principles when mapping the boosted detector to similar data-parallel architectures.
URI: http://scholarbank.nus.edu.sg/handle/10635/15790
Appears in Collections:Master's Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
Thesis_FENG_Jimin.pdf396.71 kBAdobe PDF

OPEN

NoneView/Download

Page view(s)

222
checked on Dec 11, 2017

Download(s)

259
checked on Dec 11, 2017

Google ScholarTM

Check


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