Ensembles of novel visual keywords descriptors for image categorization

Azizi Abdullah, Remco C. Veltkamp, Marco A. Wiering

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

13 Citations (Scopus)

Abstract

Object recognition systems need effective image descriptors to obtain good performance levels. Currently, the most widely used image descriptor is the SIFT descriptor that computes histograms of orientation gradients around points in an image. A possible problem of this approach is that the number of features becomes very large when a dense grid is used where the histograms are computed and combined for many different points. The current dominating solution to this problem is to use a clustering method to create a visual codebook that is exploited by an appearance based descriptor to create a histogram of visual keywords present in an image. In this paper we introduce several novel bag of visual keywords methods and compare them with the currently dominating hard bag-of-features (HBOF) approach that uses a hard assignment scheme to compute cluster frequencies. Furthermore, we combine all descriptors with a spatial pyramid and two ensemble classifiers. Experimental results on 10 and 101 classes of the Caltech-101 object database show that our novel methods significantly outperform the traditional HBOF approach and that our ensemble methods obtain state-of-the-art performance levels.

Original languageEnglish
Title of host publication11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010
Pages1206-1211
Number of pages6
DOIs
Publication statusPublished - 2010
Event11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010 - Singapore
Duration: 7 Dec 201010 Dec 2010

Other

Other11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010
CitySingapore
Period7/12/1010/12/10

Fingerprint

Object recognition
Classifiers
Object-oriented databases

ASJC Scopus subject areas

  • Artificial Intelligence
  • Control and Systems Engineering

Cite this

Abdullah, A., Veltkamp, R. C., & Wiering, M. A. (2010). Ensembles of novel visual keywords descriptors for image categorization. In 11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010 (pp. 1206-1211). [5707326] https://doi.org/10.1109/ICARCV.2010.5707326

Ensembles of novel visual keywords descriptors for image categorization. / Abdullah, Azizi; Veltkamp, Remco C.; Wiering, Marco A.

11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010. 2010. p. 1206-1211 5707326.

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

Abdullah, A, Veltkamp, RC & Wiering, MA 2010, Ensembles of novel visual keywords descriptors for image categorization. in 11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010., 5707326, pp. 1206-1211, 11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010, Singapore, 7/12/10. https://doi.org/10.1109/ICARCV.2010.5707326
Abdullah A, Veltkamp RC, Wiering MA. Ensembles of novel visual keywords descriptors for image categorization. In 11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010. 2010. p. 1206-1211. 5707326 https://doi.org/10.1109/ICARCV.2010.5707326
Abdullah, Azizi ; Veltkamp, Remco C. ; Wiering, Marco A. / Ensembles of novel visual keywords descriptors for image categorization. 11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010. 2010. pp. 1206-1211
@inproceedings{e2a234628129449790d7f973b8824bba,
title = "Ensembles of novel visual keywords descriptors for image categorization",
abstract = "Object recognition systems need effective image descriptors to obtain good performance levels. Currently, the most widely used image descriptor is the SIFT descriptor that computes histograms of orientation gradients around points in an image. A possible problem of this approach is that the number of features becomes very large when a dense grid is used where the histograms are computed and combined for many different points. The current dominating solution to this problem is to use a clustering method to create a visual codebook that is exploited by an appearance based descriptor to create a histogram of visual keywords present in an image. In this paper we introduce several novel bag of visual keywords methods and compare them with the currently dominating hard bag-of-features (HBOF) approach that uses a hard assignment scheme to compute cluster frequencies. Furthermore, we combine all descriptors with a spatial pyramid and two ensemble classifiers. Experimental results on 10 and 101 classes of the Caltech-101 object database show that our novel methods significantly outperform the traditional HBOF approach and that our ensemble methods obtain state-of-the-art performance levels.",
author = "Azizi Abdullah and Veltkamp, {Remco C.} and Wiering, {Marco A.}",
year = "2010",
doi = "10.1109/ICARCV.2010.5707326",
language = "English",
isbn = "9781424478132",
pages = "1206--1211",
booktitle = "11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010",

}

TY - GEN

T1 - Ensembles of novel visual keywords descriptors for image categorization

AU - Abdullah, Azizi

AU - Veltkamp, Remco C.

AU - Wiering, Marco A.

PY - 2010

Y1 - 2010

N2 - Object recognition systems need effective image descriptors to obtain good performance levels. Currently, the most widely used image descriptor is the SIFT descriptor that computes histograms of orientation gradients around points in an image. A possible problem of this approach is that the number of features becomes very large when a dense grid is used where the histograms are computed and combined for many different points. The current dominating solution to this problem is to use a clustering method to create a visual codebook that is exploited by an appearance based descriptor to create a histogram of visual keywords present in an image. In this paper we introduce several novel bag of visual keywords methods and compare them with the currently dominating hard bag-of-features (HBOF) approach that uses a hard assignment scheme to compute cluster frequencies. Furthermore, we combine all descriptors with a spatial pyramid and two ensemble classifiers. Experimental results on 10 and 101 classes of the Caltech-101 object database show that our novel methods significantly outperform the traditional HBOF approach and that our ensemble methods obtain state-of-the-art performance levels.

AB - Object recognition systems need effective image descriptors to obtain good performance levels. Currently, the most widely used image descriptor is the SIFT descriptor that computes histograms of orientation gradients around points in an image. A possible problem of this approach is that the number of features becomes very large when a dense grid is used where the histograms are computed and combined for many different points. The current dominating solution to this problem is to use a clustering method to create a visual codebook that is exploited by an appearance based descriptor to create a histogram of visual keywords present in an image. In this paper we introduce several novel bag of visual keywords methods and compare them with the currently dominating hard bag-of-features (HBOF) approach that uses a hard assignment scheme to compute cluster frequencies. Furthermore, we combine all descriptors with a spatial pyramid and two ensemble classifiers. Experimental results on 10 and 101 classes of the Caltech-101 object database show that our novel methods significantly outperform the traditional HBOF approach and that our ensemble methods obtain state-of-the-art performance levels.

UR - http://www.scopus.com/inward/record.url?scp=79952385806&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952385806&partnerID=8YFLogxK

U2 - 10.1109/ICARCV.2010.5707326

DO - 10.1109/ICARCV.2010.5707326

M3 - Conference contribution

SN - 9781424478132

SP - 1206

EP - 1211

BT - 11th International Conference on Control, Automation, Robotics and Vision, ICARCV 2010

ER -