Advanced

Optimizing Visual Vocabularies Using Soft Assignment Entropies

Kuang, Yubin LU ; Åström, Karl LU ; Kopp, Lars LU ; Oskarsson, Magnus LU and Byröd, Martin LU (2011) 10th Asian Conference on Computer Vision (ACCV 2010), 2010 In Lecture Notes in Computer Science 6495. p.255-268
Abstract
The state of the art for large database object retrieval in images is based on quantizing descriptors of interest points into visual words. High similarity between matching image representations (as bags of words) is based upon the assumption that matched points in the two images end up in similar words in hard assignment or in similar representations in soft assignment techniques. In this paper we study how ground truth correspondences can be used to generate better visual vocabularies. Matching of image patches can be done e.g. using deformable models or from estimating 3D geometry. For optimization of the vocabulary, we propose minimizing the entropies of soft assignment of points. We base our clustering on hierarchical k-splits. The... (More)
The state of the art for large database object retrieval in images is based on quantizing descriptors of interest points into visual words. High similarity between matching image representations (as bags of words) is based upon the assumption that matched points in the two images end up in similar words in hard assignment or in similar representations in soft assignment techniques. In this paper we study how ground truth correspondences can be used to generate better visual vocabularies. Matching of image patches can be done e.g. using deformable models or from estimating 3D geometry. For optimization of the vocabulary, we propose minimizing the entropies of soft assignment of points. We base our clustering on hierarchical k-splits. The results from our entropy based clustering are compared with hierarchical k-means. The vocabularies have been tested on real data with decreased entropy and increased true positive rate, as well as better retrieval performance. (Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
in
Lecture Notes in Computer Science
volume
6495
pages
14 pages
publisher
Springer
conference name
10th Asian Conference on Computer Vision (ACCV 2010), 2010
external identifiers
  • wos:000298617900021
  • scopus:79952500541
ISSN
1611-3349
0302-9743
ISBN
978-3-642-19281-4 (print)
978-3-642-19282-1(online)
DOI
10.1007/978-3-642-19282-1
language
English
LU publication?
yes
id
288b4328-921f-48c1-ba08-ad9a62a791be (old id 1689426)
date added to LUP
2011-06-10 22:13:59
date last changed
2017-10-08 03:06:16
@inproceedings{288b4328-921f-48c1-ba08-ad9a62a791be,
  abstract     = {The state of the art for large database object retrieval in images is based on quantizing descriptors of interest points into visual words. High similarity between matching image representations (as bags of words) is based upon the assumption that matched points in the two images end up in similar words in hard assignment or in similar representations in soft assignment techniques. In this paper we study how ground truth correspondences can be used to generate better visual vocabularies. Matching of image patches can be done e.g. using deformable models or from estimating 3D geometry. For optimization of the vocabulary, we propose minimizing the entropies of soft assignment of points. We base our clustering on hierarchical k-splits. The results from our entropy based clustering are compared with hierarchical k-means. The vocabularies have been tested on real data with decreased entropy and increased true positive rate, as well as better retrieval performance.},
  author       = {Kuang, Yubin and Åström, Karl and Kopp, Lars and Oskarsson, Magnus and Byröd, Martin},
  booktitle    = {Lecture Notes in Computer Science},
  isbn         = {978-3-642-19281-4 (print)},
  issn         = {1611-3349},
  language     = {eng},
  pages        = {255--268},
  publisher    = {Springer},
  title        = {Optimizing Visual Vocabularies Using Soft Assignment Entropies},
  url          = {http://dx.doi.org/10.1007/978-3-642-19282-1},
  volume       = {6495},
  year         = {2011},
}