On Taxonomies for Multiclass Image Categorization

Authors: Alexander Binder KlausRobert Müller Motoaki Kawanabe

Publish Date: 2011/01/15

Volume: 99, Issue: 3, Pages: 281-301

PDF Link

Abstract

We study the problem of classifying images into a given predetermined taxonomy This task can be elegantly translated into the structured learning framework However despite its power structured learning has known limits in scalability due to its high memory requirements and slow training process We propose an efficient approximation of the structured learning approach by an ensemble of local support vector machines SVMs that can be trained efficiently with standard techniques A first theoretical discussion and experiments on toydata allow to shed light onto why taxonomybased classification can outperform taxonomyfree approaches and why an appropriately combined ensemble of local SVMs might be of high practical use Further empirical results on subsets of Caltech256 and VOC2006 data indeed show that our local SVM formulation can effectively exploit the taxonomy structure and thus outperforms standard multiclass classification algorithms while it achieves on par results with taxonomybased structured algorithms at a significantly decreased computing timeThis article is published under an open access license Please check the Copyright Information section for details of this license and what reuse is permitted If your intended use exceeds what is permitted by the license or if you are unable to locate the licence and reuse information please contact the Rights and Permissions team

Keywords: