HAL - INRIA :: [inria-00633013, version 1] Aggregating local image descriptors into compact codes

inria-00633013, version 1

Aggregating local image descriptors into compact codes

Hervé Jégou () ¹, Florent Perronnin^a^, 2, Matthijs Douze ()^c^, 3, 4, Jorge Sánchez ², Patrick Pérez^b^, 5, Cordelia Schmid ()^c^, 3

IEEE Transactions on Pattern Analysis and Machine Intelligence (2011)

Abstract: This paper addresses the problem of large-scale image search. Three constraints have to be taken into account: search accuracy, efficiency, and memory usage. We first present and evaluate different ways of aggregating local image descriptors into a vector and show that the Fisher kernel achieves better performance than the reference bag-of-visual words approach for any given vector dimension. We then jointly optimize dimensionality reduction and indexing in order to obtain a precise vector comparison as well as a compact representation. The evaluation shows that the image representation can be reduced to a few dozen bytes while preserving high accuracy. Searching a 100 million image dataset takes about 250 ms on one processor core.

a – Xerox
b – Technicolor
c – INRIA
1: TEXMEX (INRIA - IRISA)
CNRS : UMR6074 – INRIA – INSA Rennes – Université de Rennes 1
2: Xerox Research Centre Europe (XRCE)
Xerox
3: LEAR (INRIA Grenoble Rhône-Alpes / LJK Laboratoire Jean Kuntzmann)
CNRS : FR71 – CNRS : UMR5527 – INRIA – Laboratoire Jean Kuntzmann – Université Joseph Fourier - Grenoble I – Institut National Polytechnique de Grenoble (INPG)
4: Service Expérimentation et Développement (SED)
INRIA
5: Technicolor R & I
Technicolor

inria-00633013, version 1
http://hal.inria.fr/inria-00633013
oai:hal.inria.fr:inria-00633013
From: Hervé Jégou <>
Submitted on: Monday, 17 October 2011 14:18:08
Updated on: Friday, 18 November 2011 10:07:51

See detailed view

Associated documents

PDF :

Export

Bibtex EndNote TEI RefWorks

<?xml version="1.0" encoding="UTF-8"?>
<listBibl xmlns="http://www.tei-c.org/ns/1.0" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML">
<listBibl type="article"><head>Articles in peer-reviewed journal</head>
<biblStruct type="article" xml:id="inria-00633013"><analytic>
<author><persName><forename>Hervé</forename><surname>Jégou</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA - IRISA</orgName>
<orgName type="team">TEXMEX</orgName>
<country>FR</country>
<orgName type="institution">CNRS : UMR6074</orgName>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">INSA Rennes</orgName>
<orgName type="institution">Université de Rennes 1</orgName>
</affiliation>
</author>
<author><persName><forename>Florent</forename><surname>Perronnin</surname></persName>
<affiliation>
<orgName type="laboratory">XRCE</orgName>
<orgName type="team">Xerox Research Centre Europe</orgName>
<country>FR</country>
<orgName type="institution">Xerox</orgName>
</affiliation>
</author>
<author><persName><forename>Matthijs</forename><surname>Douze</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA Grenoble Rhône-Alpes / LJK Laboratoire Jean Kuntzmann</orgName>
<orgName type="team">LEAR</orgName>
<country>FR</country>
<orgName type="institution">CNRS : FR71</orgName>
<orgName type="institution">CNRS : UMR5527</orgName>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">Laboratoire Jean Kuntzmann</orgName>
<orgName type="institution">Université Joseph Fourier - Grenoble I</orgName>
<orgName type="institution">Institut National Polytechnique de Grenoble (INPG)</orgName>
</affiliation>
<affiliation>
<orgName type="laboratory">SED</orgName>
<orgName type="team">Service Expérimentation et Développement</orgName>
<country>FR</country>
<orgName type="institution">INRIA</orgName>
</affiliation>
</author>
<author><persName><forename>Jorge</forename><surname>Sánchez</surname></persName>
<affiliation>
<orgName type="laboratory">XRCE</orgName>
<orgName type="team">Xerox Research Centre Europe</orgName>
<country>FR</country>
<orgName type="institution">Xerox</orgName>
</affiliation>
</author>
<author><persName><forename>Patrick</forename><surname>Pérez</surname></persName>
<affiliation>
<orgName type="team">Technicolor R & I</orgName>
<country>FR</country>
<orgName type="institution">Technicolor</orgName>
</affiliation>
</author>
<author><persName><forename>Cordelia</forename><surname>Schmid</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA Grenoble Rhône-Alpes / LJK Laboratoire Jean Kuntzmann</orgName>
<orgName type="team">LEAR</orgName>
<country>FR</country>
<orgName type="institution">CNRS : FR71</orgName>
<orgName type="institution">CNRS : UMR5527</orgName>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">Laboratoire Jean Kuntzmann</orgName>
<orgName type="institution">Université Joseph Fourier - Grenoble I</orgName>
<orgName type="institution">Institut National Polytechnique de Grenoble (INPG)</orgName>
</affiliation>
</author>
<title type="main" level="a">Aggregating local image descriptors into compact codes</title></analytic><monogr><title level="j" type="main">IEEE Transactions on Pattern Analysis and Machine Intelligence</title><imprint><publisher>IEEE</publisher><biblScope type="publicationDate">2011-00-00</biblScope></imprint></monogr><idno type="HALid">http://hal.inria.fr/inria-00633013</idno><idno type="HALFile">http://hal.inria.fr/inria-00633013/PDF/jegou_aggregate.pdf</idno></biblStruct>
</listBibl>
</listBibl>