Small codes and large image databases for recognition

Antonio Torralba, Robert Fergus, Yair Weiss

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

Abstract

The Internet contains billions of images, freely available online. Methods for efficiently searching this incredibly rich resource are vital for a large number of applications. These include object recognition [2], computer graphics [11, 27], personal photo collections, online image search tools. In this paper, our goal is to develop efficient image search and scene matching techniques that are not only fast, but also require very little memory, enabling their use on standard hardware or even on handheld devices. Our approach uses recently developed machine learning techniques to convert the Gist descriptor (a real valued vector that describes orientation energies at different scales and orientations within an image) to a compact binary code, with a few hundred bits per image. Using our scheme, it is possible to perform real-time searches with millions from the Internet using a single large PC and obtain recognition results comparable to the full descriptor. Using our codes on high quality labeled images from the LabelMe database gives surprisingly powerful recognition results using simple nearest neighbor techniques.

Original languageEnglish (US)
Title of host publication26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR
DOIs
StatePublished - 2008
Event26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR - Anchorage, AK, United States
Duration: Jun 23 2008Jun 28 2008

Other

Other26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR
CountryUnited States
CityAnchorage, AK
Period6/23/086/28/08

Fingerprint

Internet
Binary codes
Object recognition
Computer graphics
Computer hardware
Image quality
Learning systems
Data storage equipment

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Control and Systems Engineering

Cite this

Torralba, A., Fergus, R., & Weiss, Y. (2008). Small codes and large image databases for recognition. In 26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR [4587633] https://doi.org/10.1109/CVPR.2008.4587633

Small codes and large image databases for recognition. / Torralba, Antonio; Fergus, Robert; Weiss, Yair.

26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR. 2008. 4587633.

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

Torralba, A, Fergus, R & Weiss, Y 2008, Small codes and large image databases for recognition. in 26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR., 4587633, 26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR, Anchorage, AK, United States, 6/23/08. https://doi.org/10.1109/CVPR.2008.4587633
Torralba A, Fergus R, Weiss Y. Small codes and large image databases for recognition. In 26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR. 2008. 4587633 https://doi.org/10.1109/CVPR.2008.4587633
Torralba, Antonio ; Fergus, Robert ; Weiss, Yair. / Small codes and large image databases for recognition. 26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR. 2008.
@inproceedings{133e90604c374fb5a85d926837eafd0f,
title = "Small codes and large image databases for recognition",
abstract = "The Internet contains billions of images, freely available online. Methods for efficiently searching this incredibly rich resource are vital for a large number of applications. These include object recognition [2], computer graphics [11, 27], personal photo collections, online image search tools. In this paper, our goal is to develop efficient image search and scene matching techniques that are not only fast, but also require very little memory, enabling their use on standard hardware or even on handheld devices. Our approach uses recently developed machine learning techniques to convert the Gist descriptor (a real valued vector that describes orientation energies at different scales and orientations within an image) to a compact binary code, with a few hundred bits per image. Using our scheme, it is possible to perform real-time searches with millions from the Internet using a single large PC and obtain recognition results comparable to the full descriptor. Using our codes on high quality labeled images from the LabelMe database gives surprisingly powerful recognition results using simple nearest neighbor techniques.",
author = "Antonio Torralba and Robert Fergus and Yair Weiss",
year = "2008",
doi = "10.1109/CVPR.2008.4587633",
language = "English (US)",
isbn = "9781424422432",
booktitle = "26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR",

}

TY - GEN

T1 - Small codes and large image databases for recognition

AU - Torralba, Antonio

AU - Fergus, Robert

AU - Weiss, Yair

PY - 2008

Y1 - 2008

N2 - The Internet contains billions of images, freely available online. Methods for efficiently searching this incredibly rich resource are vital for a large number of applications. These include object recognition [2], computer graphics [11, 27], personal photo collections, online image search tools. In this paper, our goal is to develop efficient image search and scene matching techniques that are not only fast, but also require very little memory, enabling their use on standard hardware or even on handheld devices. Our approach uses recently developed machine learning techniques to convert the Gist descriptor (a real valued vector that describes orientation energies at different scales and orientations within an image) to a compact binary code, with a few hundred bits per image. Using our scheme, it is possible to perform real-time searches with millions from the Internet using a single large PC and obtain recognition results comparable to the full descriptor. Using our codes on high quality labeled images from the LabelMe database gives surprisingly powerful recognition results using simple nearest neighbor techniques.

AB - The Internet contains billions of images, freely available online. Methods for efficiently searching this incredibly rich resource are vital for a large number of applications. These include object recognition [2], computer graphics [11, 27], personal photo collections, online image search tools. In this paper, our goal is to develop efficient image search and scene matching techniques that are not only fast, but also require very little memory, enabling their use on standard hardware or even on handheld devices. Our approach uses recently developed machine learning techniques to convert the Gist descriptor (a real valued vector that describes orientation energies at different scales and orientations within an image) to a compact binary code, with a few hundred bits per image. Using our scheme, it is possible to perform real-time searches with millions from the Internet using a single large PC and obtain recognition results comparable to the full descriptor. Using our codes on high quality labeled images from the LabelMe database gives surprisingly powerful recognition results using simple nearest neighbor techniques.

UR - http://www.scopus.com/inward/record.url?scp=51949119257&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=51949119257&partnerID=8YFLogxK

U2 - 10.1109/CVPR.2008.4587633

DO - 10.1109/CVPR.2008.4587633

M3 - Conference contribution

AN - SCOPUS:51949119257

SN - 9781424422432

BT - 26th IEEE Conference on Computer Vision and Pattern Recognition, CVPR

ER -