Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Object Recognition using Geometric Properties and a variant of Boosting1)

Summary: Object Recognition using Geometric Properties and a
variant of Boosting1)
Martin Antenreiter and Peter Auer
Chair of Information Technology (CiT), University of Leoben,
{martin.antenreiter, auer}@unileoben.ac.at,
This paper describes an approach for learning object descriptions as combinations of simple
features using labeled still images. The contribution of this paper is a new method for con-
structing geometric relations of simple features with the LPBoost algorithm. A full search for
relevant geometric relations between simple features is rather impossible because of the compu-
tation time required. We therefore use a greedy search strategy. In comparison, our recognition
results are better than results using related approaches.
1 Introduction
Object detection and object recognition are important subtasks in Hide&Seek scenarios. In
this work we investigate improved mechanisms for learning object descriptions from exam-
ples. In particular, we are interested in the description of objects by combinations of simple
features based on the boosting technology [3, 1]. Whereas boosting originally delivers a linear
combination of simple weak hypotheses, our extension delivers a linear combination of more


Source: Auer, Peter - Department Mathematik und Informationstechnologie, Montanuniversitšt Leoben


Collections: Computer Technologies and Information Sciences