A Study on an Effective Feature Selection Method Using Hog-Family Feature for Human Detection
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F44555601%3A13440%2F14%3A43886126" target="_blank" >RIV/44555601:13440/14:43886126 - isvavai.cz</a>
Result on the web
<a href="http://www.sersc.org/journals/IJMUE/vol9_no12_2014/19.pdf" target="_blank" >http://www.sersc.org/journals/IJMUE/vol9_no12_2014/19.pdf</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.14257/ijmue.2014.9.12.19" target="_blank" >10.14257/ijmue.2014.9.12.19</a>
Alternative languages
Result language
angličtina
Original language name
A Study on an Effective Feature Selection Method Using Hog-Family Feature for Human Detection
Original language description
It is the important that Support Vector Machine (SVM) is the powerful learning machines and has been applied to varying task with generally acceptable performance. The SVM success for classification tasks in one domain is affected by features that it represents the instance of specific class. The representative and discriminative features that they are given, SVM learning is going to provide better generalization and consequently that we are able to obtain good classifier. In this paper, we define the problem of feature choices for tasks of human detections and measure the performance of each feature. And also we consider HOG-family feature to study an effective feature selection method. Finally we proposed the multi-scale HOG as a NEW family member inthis feature group. In addition we also combine SVM with Principal Component Analysis (PCA) to reduce dimension of features and enhance the evaluation speed while retaining most of discriminative feature vectors.
Czech name
—
Czech description
—
Classification
Type
J<sub>x</sub> - Unclassified - Peer-reviewed scientific article (Jimp, Jsc and Jost)
CEP classification
IN - Informatics
OECD FORD branch
—
Result continuities
Project
—
Continuities
N - Vyzkumna aktivita podporovana z neverejnych zdroju
Others
Publication year
2014
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Name of the periodical
International Journal of Multimedia and Ubiquitous Engineering
ISSN
1975-0080
e-ISSN
—
Volume of the periodical
9
Issue of the periodical within the volume
12
Country of publishing house
KR - KOREA, REPUBLIC OF
Number of pages
9
Pages from-to
203-212
UT code for WoS article
—
EID of the result in the Scopus database
—