Learning of a Robusted Nearest Neighbor Classifier Using Multiple Training Data

Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26220%2F16%3APU120033" target="_blank" >RIV/00216305:26220/16:PU120033 - isvavai.cz</a>
Result on the web
<a href="https://ieeexplore.ieee.org/document/7502701" target="_blank" >https://ieeexplore.ieee.org/document/7502701</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1109/IWSSIP.2016.7502701" target="_blank" >10.1109/IWSSIP.2016.7502701</a>

Result language
angličtina
Original language name
Learning of a Robusted Nearest Neighbor Classifier Using Multiple Training Data
Original language description
This paper deals with the application of face recognition in surveillance CCTV systems and effective usage of so called recognition clues. These clues are enrollment of multiple training face images and their usages in classifier training and real-time management of template database. A survey on classifiers from perspective of practical application is given resulting in the defense of nearest neighbor based classifiers. They are competitive with state of the art classifiers and are suitable for practical application. Template creation using multiple training face images and enhancement of NN-based classifier performance is achieved by novel approach. It consist of quantile interval method for template creation and robusted NNbased classifier using spatial templates with soft boundaries. We evaluate proposed recognition framework on highly representative IFaViD dataset. Proposed framework outperforms state of the art approaches.
Czech name
—
Czech description
—

Project
<a href="/en/project/LO1401" target="_blank" >LO1401: Interdisciplinary Research of Wireless Technologies</a><br>
Continuities
P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Publication year
2016
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection
Proceedings The 23rd International Conference on Systems, Signals and Image Processing
ISBN
978-1-4673-9554-0
ISSN
—
e-ISSN
—
Number of pages
4
Pages from-to
47-50
Publisher name
Neuveden
Place of publication
Bratislava
Event location
Bratislava
Event date
May 23, 2016
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
000389830400006

Similar results(10)