A Performance Evaluation of Exact and Approximate Match Kernels for Object Recognition
Local features have repeatedly shown their effectiveness for object recognition during the last years, and they have consequently become the preferred descriptor for this type of problems. The solution of the correspondence problem is traditionally approached with exact or approximate techniques. I...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
Computer Vision Center Press
2010-02-01
|
Series: | ELCVIA Electronic Letters on Computer Vision and Image Analysis |
Subjects: | |
Online Access: | https://elcvia.cvc.uab.es/article/view/350 |
id |
doaj-e2008c32c6764ed4954166a331f48e81 |
---|---|
record_format |
Article |
spelling |
doaj-e2008c32c6764ed4954166a331f48e812021-09-18T12:40:05ZengComputer Vision Center PressELCVIA Electronic Letters on Computer Vision and Image Analysis1577-50972010-02-018310.5565/rev/elcvia.350176A Performance Evaluation of Exact and Approximate Match Kernels for Object RecognitionBarbara CaputoLuo Jie Local features have repeatedly shown their effectiveness for object recognition during the last years, and they have consequently become the preferred descriptor for this type of problems. The solution of the correspondence problem is traditionally approached with exact or approximate techniques. In this paper we are interested in methods that solve the correspondence problem via the definition of a kernel function that makes it possible to use local features as input to a support vector machine. We single out the match kernel, an exact approach, and the pyramid match kernel, that uses instead an approximate strategy. We present a thorough experimental evaluation of the two methods on three different databases. Results show that the exact method performs consistently better than the approximate one, especially for the object identification task, when training on a decreasing number of images. Based on this findings and on the computational cost of each approach, we suggest some criteria for choosing between the two kernels given the application at hand. https://elcvia.cvc.uab.es/article/view/350Object Description and RecognitionSupport Vector Machines and kernels |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Barbara Caputo Luo Jie |
spellingShingle |
Barbara Caputo Luo Jie A Performance Evaluation of Exact and Approximate Match Kernels for Object Recognition ELCVIA Electronic Letters on Computer Vision and Image Analysis Object Description and Recognition Support Vector Machines and kernels |
author_facet |
Barbara Caputo Luo Jie |
author_sort |
Barbara Caputo |
title |
A Performance Evaluation of Exact and Approximate Match Kernels for Object Recognition |
title_short |
A Performance Evaluation of Exact and Approximate Match Kernels for Object Recognition |
title_full |
A Performance Evaluation of Exact and Approximate Match Kernels for Object Recognition |
title_fullStr |
A Performance Evaluation of Exact and Approximate Match Kernels for Object Recognition |
title_full_unstemmed |
A Performance Evaluation of Exact and Approximate Match Kernels for Object Recognition |
title_sort |
performance evaluation of exact and approximate match kernels for object recognition |
publisher |
Computer Vision Center Press |
series |
ELCVIA Electronic Letters on Computer Vision and Image Analysis |
issn |
1577-5097 |
publishDate |
2010-02-01 |
description |
Local features have repeatedly shown their effectiveness for object recognition during the last years, and they have consequently become the preferred descriptor for this type of problems. The solution of the correspondence problem is traditionally approached with exact or approximate techniques. In this paper we are interested in methods that solve the correspondence problem via the definition of a kernel function that makes it possible to use local features as input to a support vector machine. We single out the match kernel, an exact approach, and the pyramid match kernel, that uses instead an approximate strategy. We present a thorough experimental evaluation of the two methods on three different databases. Results show that the exact method performs consistently better than the approximate one, especially for the object identification task, when training on a decreasing number of images. Based on this findings and on the computational cost of each approach, we suggest some criteria for choosing between the two kernels given the application at hand.
|
topic |
Object Description and Recognition Support Vector Machines and kernels |
url |
https://elcvia.cvc.uab.es/article/view/350 |
work_keys_str_mv |
AT barbaracaputo aperformanceevaluationofexactandapproximatematchkernelsforobjectrecognition AT luojie aperformanceevaluationofexactandapproximatematchkernelsforobjectrecognition AT barbaracaputo performanceevaluationofexactandapproximatematchkernelsforobjectrecognition AT luojie performanceevaluationofexactandapproximatematchkernelsforobjectrecognition |
_version_ |
1717376937157984256 |