Comparative object similarity for improved recognition with few or no examples

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Learning models for recognizing objects with few or no training examples is important, due to the intrinsic long-tailed distribution of objects in the real world. In this paper, we propose an approach to use comparative object similarity. The key insight is that: given a set of object categories which are similar and a set of categories which are dissimilar, a good object model should respond more strongly to examples from similar categories than to examples from dissimilar categories. We develop a regularized kernel machine algorithm to use this category dependent similarity regularization. Our experiments on hundreds of categories show that our method can make significant improvement, especially for categories with no examples.

Original languageEnglish (US)
Title of host publication2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010
Pages3525-3532
Number of pages8
DOIs
StatePublished - Aug 31 2010
Event2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010 - San Francisco, CA, United States
Duration: Jun 13 2010Jun 18 2010

Publication series

NameProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
ISSN (Print)1063-6919

Other

Other2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010
CountryUnited States
CitySan Francisco, CA
Period6/13/106/18/10

Fingerprint

Experiments

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition

Cite this

Wang, G., Forsyth, D. A., & Hoiem, D. W. (2010). Comparative object similarity for improved recognition with few or no examples. In 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010 (pp. 3525-3532). [5539955] (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition). https://doi.org/10.1109/CVPR.2010.5539955

Comparative object similarity for improved recognition with few or no examples. / Wang, Gang; Forsyth, David Alexander; Hoiem, Derek W.

2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010. 2010. p. 3525-3532 5539955 (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Wang, G, Forsyth, DA & Hoiem, DW 2010, Comparative object similarity for improved recognition with few or no examples. in 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010., 5539955, Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 3525-3532, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010, San Francisco, CA, United States, 6/13/10. https://doi.org/10.1109/CVPR.2010.5539955
Wang G, Forsyth DA, Hoiem DW. Comparative object similarity for improved recognition with few or no examples. In 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010. 2010. p. 3525-3532. 5539955. (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition). https://doi.org/10.1109/CVPR.2010.5539955
Wang, Gang ; Forsyth, David Alexander ; Hoiem, Derek W. / Comparative object similarity for improved recognition with few or no examples. 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010. 2010. pp. 3525-3532 (Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition).
@inproceedings{f4359e62ef1a4be8a9cec0a87545c733,
title = "Comparative object similarity for improved recognition with few or no examples",
abstract = "Learning models for recognizing objects with few or no training examples is important, due to the intrinsic long-tailed distribution of objects in the real world. In this paper, we propose an approach to use comparative object similarity. The key insight is that: given a set of object categories which are similar and a set of categories which are dissimilar, a good object model should respond more strongly to examples from similar categories than to examples from dissimilar categories. We develop a regularized kernel machine algorithm to use this category dependent similarity regularization. Our experiments on hundreds of categories show that our method can make significant improvement, especially for categories with no examples.",
author = "Gang Wang and Forsyth, {David Alexander} and Hoiem, {Derek W}",
year = "2010",
month = "8",
day = "31",
doi = "10.1109/CVPR.2010.5539955",
language = "English (US)",
isbn = "9781424469840",
series = "Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition",
pages = "3525--3532",
booktitle = "2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010",

}

TY - GEN

T1 - Comparative object similarity for improved recognition with few or no examples

AU - Wang, Gang

AU - Forsyth, David Alexander

AU - Hoiem, Derek W

PY - 2010/8/31

Y1 - 2010/8/31

N2 - Learning models for recognizing objects with few or no training examples is important, due to the intrinsic long-tailed distribution of objects in the real world. In this paper, we propose an approach to use comparative object similarity. The key insight is that: given a set of object categories which are similar and a set of categories which are dissimilar, a good object model should respond more strongly to examples from similar categories than to examples from dissimilar categories. We develop a regularized kernel machine algorithm to use this category dependent similarity regularization. Our experiments on hundreds of categories show that our method can make significant improvement, especially for categories with no examples.

AB - Learning models for recognizing objects with few or no training examples is important, due to the intrinsic long-tailed distribution of objects in the real world. In this paper, we propose an approach to use comparative object similarity. The key insight is that: given a set of object categories which are similar and a set of categories which are dissimilar, a good object model should respond more strongly to examples from similar categories than to examples from dissimilar categories. We develop a regularized kernel machine algorithm to use this category dependent similarity regularization. Our experiments on hundreds of categories show that our method can make significant improvement, especially for categories with no examples.

UR - http://www.scopus.com/inward/record.url?scp=77955997493&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77955997493&partnerID=8YFLogxK

U2 - 10.1109/CVPR.2010.5539955

DO - 10.1109/CVPR.2010.5539955

M3 - Conference contribution

AN - SCOPUS:77955997493

SN - 9781424469840

T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

SP - 3525

EP - 3532

BT - 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2010

ER -