Publication detail

Grammatical Evolution as a Learning Process for Multiclass Object Detection

ŠKORPIL, V. LÝSEK, J. MOTYČKA, A. CEPL, M. ENDRLE, P.

Original Title

Grammatical Evolution as a Learning Process for Multiclass Object Detection

English Title

Grammatical Evolution as a Learning Process for Multiclass Object Detection

Type

conference paper

Language

en

Original Abstract

In this article we describe the usage of grammatical evolution as a learning process. Evolved programs are used as a sliding window to recognize different objects in an image. Programs are trained on a set of greyscale images with object locations and object class identification. Best individuals are tested on different sets of images of simulated technological scenes with objects of different classes and their performance is reported. Grammatical evolution uses rewriting rules to translate chromosome into a tree structure programs.

English abstract

In this article we describe the usage of grammatical evolution as a learning process. Evolved programs are used as a sliding window to recognize different objects in an image. Programs are trained on a set of greyscale images with object locations and object class identification. Best individuals are tested on different sets of images of simulated technological scenes with objects of different classes and their performance is reported. Grammatical evolution uses rewriting rules to translate chromosome into a tree structure programs.

Keywords

grammatical evolution, learning, object recognition

RIV year

2012

Released

01.07.2012

ISBN

978-1-61804-108-1

Book

Recent Researches in Communications and Computers

Pages from

101

Pages to

105

Pages count

5

Documents

BibTex


@inproceedings{BUT93427,
  author="Vladislav {Škorpil} and Jiří {Lýsek} and Arnošt {Motyčka} and Miroslav {Cepl} and Pavel {Endrle}",
  title="Grammatical Evolution as a Learning Process for Multiclass Object Detection",
  annote="In this article we describe the usage of grammatical evolution as a learning process. Evolved programs are used as a sliding window to recognize different objects in an image. Programs are trained on a set of greyscale images with object locations and object class identification. Best individuals are tested on different sets of images of simulated technological scenes with objects of different classes and their performance is reported. Grammatical evolution uses rewriting rules to translate chromosome into a tree structure programs.",
  booktitle="Recent Researches in Communications and Computers",
  chapter="93427",
  howpublished="print",
  year="2012",
  month="july",
  pages="101--105",
  type="conference paper"
}