Publication Details

Evolutionary Approximation of Gradient Orientation Module in HOG-based Human Detection System

WIGLASZ, M.; SEKANINA, L. Evolutionary Approximation of Gradient Orientation Module in HOG-based Human Detection System. In 2017 IEEE Global Conference on Signal and Information Processing GlobalSIP 2017. Montreal: IEEE Signal Processing Society, 2017. p. 1300-1304. ISBN: 978-1-5090-5989-8.
Czech title
Evoluční aproximace výpočtu orientace gradientu v systému detekce osob založeném na HOG
Type
conference paper
Language
English
Authors
Wiglasz Michal, Ing.
Sekanina Lukáš, prof. Ing., Ph.D. (DCSY)
Keywords

Functional approximation, Cartesian genetic programming, Histogram of oriented gradients

Abstract

The histogram of oriented gradients (HOG) feature extraction is a computer vision method widely used in embedded systems for detection of objects such as pedestrians. We used Cartesian genetic programming (CGP) to exploit the error resilience in the HOG algorithm. We evolved new approximate implementations of the arctan function, which is typically employed to compute the gradient orientations. When the best evolved approximations are integrated into the SW implementation of the HOG algorithm, not only the execution time, but also the classification accuracy was improved in comparison with the accurate implementation and the state-of-the-art approximate implementations.

Published
2017
Pages
1300–1304
Proceedings
2017 IEEE Global Conference on Signal and Information Processing GlobalSIP 2017
ISBN
978-1-5090-5989-8
Publisher
IEEE Signal Processing Society
Place
Montreal
DOI
UT WoS
000450053100257
EID Scopus
BibTeX
@inproceedings{BUT144438,
  author="Michal {Wiglasz} and Lukáš {Sekanina}",
  title="Evolutionary Approximation of Gradient Orientation Module in HOG-based Human Detection System",
  booktitle="2017 IEEE Global Conference on Signal and Information Processing GlobalSIP 2017",
  year="2017",
  pages="1300--1304",
  publisher="IEEE Signal Processing Society",
  address="Montreal",
  doi="10.1109/GlobalSIP.2017.8309171",
  isbn="978-1-5090-5989-8",
  url="https://www.fit.vut.cz/research/publication/11441/"
}
Back to top