Conference paper

ŘEZNÍČEK Ivo. Object identification in image using global low-level features. In: Digital Technologies International Workshop 2008. Žilina: University of Žilina, 2008, p. 4. ISBN 978-80-8070-953-2.
Publication language:english
Original title:Object identification in image using global low-level features
Title (cs):Identifikace objektu v obraze při použití globálních nízkoúrovňových příznakových vektorů
Pages:4
Proceedings:Digital Technologies International Workshop 2008
Conference:Digital Technologies 2008
Series:Digital Technologies International Workshop 2008
Place:Žilina, SK
Year:2008
ISBN:978-80-8070-953-2
Publisher:University of Žilina
Keywords
global features, Support Vector Machine, feature extraction, training, libsvm, boost library, Sun Grid Engine, TRECVID.
Annotation
Paper describes system for object or situation detection in image using global low-level features. First, block diagram of system for optimal classifier creation is described. Next, all blocks are more detailed described. This approach were used also in Trecvid 2008 evaluation in task HLF extraction.
Abstract
This paper introduces a method of object recognition in image or video sequence using low-level features and powerful machine learning techniques. The goal of the method is to identify an object in unknown images or video. For this purpose optimal svm detector must be created, for which a good dataset is needed
with desired images to train this detector and for computing it's accuracy.
BibTeX:
@INPROCEEDINGS{
   author = {Ivo {\v{R}}ezn{\'{i}}{\v{c}}ek},
   title = {Object identification in image using global
	low-level features},
   pages = 4,
   booktitle = {Digital Technologies International Workshop 2008},
   series = {Digital Technologies International Workshop 2008},
   year = 2008,
   location = {{\v{Z}}ilina, SK},
   publisher = {University of {\v{Z}}ilina},
   ISBN = {978-80-8070-953-2},
   language = {english},
   url = {http://www.fit.vutbr.cz/research/view_pub.php?id=8839}
}

Your IPv4 address: 34.238.189.171
Switch to https