Efficient Classification of Images with Taxonomies

Research output: Contributions to collected editions/worksArticle in conference proceedingsResearchpeer-review

Authors

We study the problem of classifying images into a given, pre-determined taxonomy. The task can be elegantly translated into the structured learning framework. Structured learning, however, is known for its memory consuming and slow training processes. The contribution of our paper is twofold: Firstly, we propose an e.cient decomposition of the structured learning approach into an equivalent ensemble of local support vector machines (SVMs) which can be trained with standard techniques. Secondly, we combine the local SVMs to a global model by re-incorporating the taxonomy into the training process. Our empirical results on Caltech256 and VOC2006 data show that our local-global SVM effectively exploits the structure of the taxonomy and outperforms multi-class classification approaches.
Original languageEnglish
Title of host publicationComputer Vision, ACCV 2009 - 9th Asian Conference on Computer Vision, Revised Selected Papers : Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number of pages12
Place of PublicationBerlin, Heidelberg
PublisherSpringer
Publication date2010
Pages351-362
ISBN (print)978-3-642-12296-5
ISBN (electronic)978-3-642-12297-2
DOIs
Publication statusPublished - 2010
Externally publishedYes
Event9th Asian Conference on Computer Vision - ACCV 2009 - Xi'an, China
Duration: 23.09.200927.09.2009
Conference number: 9

    Research areas

  • Informatics - Empirical results, Global models, Local support, Local-global, Multi-class classification, Structured learning, Training process
  • Business informatics