Unsupervised writer adaptation of whole-word HMMs with application to word-spotting
Identifieur interne : 007426 ( Main/Exploration ); précédent : 007425; suivant : 007427Unsupervised writer adaptation of whole-word HMMs with application to word-spotting
Auteurs : José A. Rodriguez-Serrano [France, Espagne] ; Florent Perronnin [France] ; Gemma Sanchez [Espagne] ; Josep Llados [Espagne]Source :
- Pattern recognition letters [ 0167-8655 ] ; 2010.
Descripteurs français
- Pascal (Inist)
- Wicri :
- topic : Méthode statistique.
English descriptors
- KwdEn :
Abstract
In this paper we propose a novel approach for writer adaptation in a handwritten word-spotting task. The method exploits the fact that the semi-continuous hidden Markov model separates the word model parameters into (i) a codebook of shapes and (ii) a set of word-specific parameters. Our main contribution is to employ this property to derive writer-specific word models by statistically adapting an initial universal codebook to each document. This process is unsupervised and does not even require the appearance of the keyword(s) in the searched document. Experimental results show an increase in performance when this adaptation technique is applied. To the best of our knowledge, this is the first work dealing with adaptation for word-spotting. The preliminary version of this paper obtained an IBM Best Student Paper Award at the 19th International Conference on Pattern Recognition.
Affiliations:
- Espagne, France
- Auvergne-Rhône-Alpes, Catalogne, Rhône-Alpes
- Barcelone, Meylan
- Université autonome de Barcelone
Links toward previous steps (curation, corpus...)
- to stream PascalFrancis, to step Corpus: 002709
- to stream PascalFrancis, to step Curation: 003877
- to stream PascalFrancis, to step Checkpoint: 001E98
- to stream Main, to step Merge: 007966
- to stream Main, to step Curation: 007426
Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" level="a">Unsupervised writer adaptation of whole-word HMMs with application to word-spotting</title>
<author><name sortKey="Rodriguez Serrano, Jose A" sort="Rodriguez Serrano, Jose A" uniqKey="Rodriguez Serrano J" first="José A." last="Rodriguez-Serrano">José A. Rodriguez-Serrano</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Textual and Visual Pattern Analysis, Xerox Research Centre Europe (XRCE)</s1>
<s2>38240 Meylan</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>France</country>
<wicri:noRegion>38240 Meylan</wicri:noRegion>
<wicri:noRegion>Xerox Research Centre Europe (XRCE)</wicri:noRegion>
<wicri:noRegion>38240 Meylan</wicri:noRegion>
</affiliation>
<affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Centre de Visió per Computador (CVC), Universitat Autònoma de Barcelona</s1>
<s2>08193 Bellaterra</s2>
<s3>ESP</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>Espagne</country>
<placeName><region nuts="2" type="communauté">Catalogne</region>
<settlement type="city">Barcelone</settlement>
</placeName>
<orgName type="university">Université autonome de Barcelone</orgName>
</affiliation>
</author>
<author><name sortKey="Perronnin, Florent" sort="Perronnin, Florent" uniqKey="Perronnin F" first="Florent" last="Perronnin">Florent Perronnin</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Textual and Visual Pattern Analysis, Xerox Research Centre Europe (XRCE)</s1>
<s2>38240 Meylan</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>France</country>
<wicri:noRegion>38240 Meylan</wicri:noRegion>
<wicri:noRegion>Xerox Research Centre Europe (XRCE)</wicri:noRegion>
<wicri:noRegion>38240 Meylan</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Sanchez, Gemma" sort="Sanchez, Gemma" uniqKey="Sanchez G" first="Gemma" last="Sanchez">Gemma Sanchez</name>
<affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Centre de Visió per Computador (CVC), Universitat Autònoma de Barcelona</s1>
<s2>08193 Bellaterra</s2>
<s3>ESP</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>Espagne</country>
<placeName><region nuts="2" type="communauté">Catalogne</region>
<settlement type="city">Barcelone</settlement>
</placeName>
<orgName type="university">Université autonome de Barcelone</orgName>
</affiliation>
</author>
<author><name sortKey="Llados, Josep" sort="Llados, Josep" uniqKey="Llados J" first="Josep" last="Llados">Josep Llados</name>
<affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Centre de Visió per Computador (CVC), Universitat Autònoma de Barcelona</s1>
<s2>08193 Bellaterra</s2>
<s3>ESP</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>Espagne</country>
<placeName><region nuts="2" type="communauté">Catalogne</region>
<settlement type="city">Barcelone</settlement>
</placeName>
<orgName type="university">Université autonome de Barcelone</orgName>
</affiliation>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">INIST</idno>
<idno type="inist">10-0223708</idno>
<date when="2010">2010</date>
<idno type="stanalyst">PASCAL 10-0223708 INIST</idno>
<idno type="RBID">Pascal:10-0223708</idno>
<idno type="wicri:Area/PascalFrancis/Corpus">002709</idno>
<idno type="wicri:Area/PascalFrancis/Curation">003877</idno>
<idno type="wicri:Area/PascalFrancis/Checkpoint">001E98</idno>
<idno type="wicri:explorRef" wicri:stream="PascalFrancis" wicri:step="Checkpoint">001E98</idno>
<idno type="wicri:doubleKey">0167-8655:2010:Rodriguez Serrano J:unsupervised:writer:adaptation</idno>
<idno type="wicri:Area/Main/Merge">007966</idno>
<idno type="wicri:Area/Main/Curation">007426</idno>
<idno type="wicri:Area/Main/Exploration">007426</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="en" level="a">Unsupervised writer adaptation of whole-word HMMs with application to word-spotting</title>
<author><name sortKey="Rodriguez Serrano, Jose A" sort="Rodriguez Serrano, Jose A" uniqKey="Rodriguez Serrano J" first="José A." last="Rodriguez-Serrano">José A. Rodriguez-Serrano</name>
<affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Textual and Visual Pattern Analysis, Xerox Research Centre Europe (XRCE)</s1>
<s2>38240 Meylan</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>France</country>
<placeName><region type="region" nuts="2">Auvergne-Rhône-Alpes</region>
<region type="old region" nuts="2">Rhône-Alpes</region>
<settlement type="city">Meylan</settlement>
</placeName>
</affiliation>
<affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Centre de Visió per Computador (CVC), Universitat Autònoma de Barcelona</s1>
<s2>08193 Bellaterra</s2>
<s3>ESP</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>Espagne</country>
<placeName><region nuts="2" type="communauté">Catalogne</region>
<settlement type="city">Barcelone</settlement>
</placeName>
<orgName type="university">Université autonome de Barcelone</orgName>
</affiliation>
</author>
<author><name sortKey="Perronnin, Florent" sort="Perronnin, Florent" uniqKey="Perronnin F" first="Florent" last="Perronnin">Florent Perronnin</name>
<affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Textual and Visual Pattern Analysis, Xerox Research Centre Europe (XRCE)</s1>
<s2>38240 Meylan</s2>
<s3>FRA</s3>
<sZ>1 aut.</sZ>
<sZ>2 aut.</sZ>
</inist:fA14>
<country>France</country>
<wicri:noRegion>38240 Meylan</wicri:noRegion>
<wicri:noRegion>Xerox Research Centre Europe (XRCE)</wicri:noRegion>
<wicri:noRegion>38240 Meylan</wicri:noRegion>
</affiliation>
</author>
<author><name sortKey="Sanchez, Gemma" sort="Sanchez, Gemma" uniqKey="Sanchez G" first="Gemma" last="Sanchez">Gemma Sanchez</name>
<affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Centre de Visió per Computador (CVC), Universitat Autònoma de Barcelona</s1>
<s2>08193 Bellaterra</s2>
<s3>ESP</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>Espagne</country>
<placeName><region nuts="2" type="communauté">Catalogne</region>
<settlement type="city">Barcelone</settlement>
</placeName>
<orgName type="university">Université autonome de Barcelone</orgName>
</affiliation>
</author>
<author><name sortKey="Llados, Josep" sort="Llados, Josep" uniqKey="Llados J" first="Josep" last="Llados">Josep Llados</name>
<affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Centre de Visió per Computador (CVC), Universitat Autònoma de Barcelona</s1>
<s2>08193 Bellaterra</s2>
<s3>ESP</s3>
<sZ>1 aut.</sZ>
<sZ>3 aut.</sZ>
<sZ>4 aut.</sZ>
</inist:fA14>
<country>Espagne</country>
<placeName><region nuts="2" type="communauté">Catalogne</region>
<settlement type="city">Barcelone</settlement>
</placeName>
<orgName type="university">Université autonome de Barcelone</orgName>
</affiliation>
</author>
</analytic>
<series><title level="j" type="main">Pattern recognition letters</title>
<title level="j" type="abbreviated">Pattern recogn. lett.</title>
<idno type="ISSN">0167-8655</idno>
<imprint><date when="2010">2010</date>
</imprint>
</series>
</biblStruct>
</sourceDesc>
<seriesStmt><title level="j" type="main">Pattern recognition letters</title>
<title level="j" type="abbreviated">Pattern recogn. lett.</title>
<idno type="ISSN">0167-8655</idno>
</seriesStmt>
</fileDesc>
<profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Codebook</term>
<term>Document analysis</term>
<term>Handwriting recognition</term>
<term>Hidden Markov models</term>
<term>Keyword</term>
<term>Manuscript character</term>
<term>Pattern recognition</term>
<term>Performance evaluation</term>
<term>Probabilistic approach</term>
<term>Semimarkovian process</term>
<term>Signal classification</term>
<term>Statistical method</term>
<term>Unsupervised classification</term>
</keywords>
<keywords scheme="Pascal" xml:lang="fr"><term>Classification non supervisée</term>
<term>Modèle Markov variable cachée</term>
<term>Caractère manuscrit</term>
<term>Processus semi markovien</term>
<term>Table codage</term>
<term>Méthode statistique</term>
<term>Mot clé</term>
<term>Evaluation performance</term>
<term>Reconnaissance forme</term>
<term>Reconnaissance écriture</term>
<term>Analyse documentaire</term>
<term>Classification signal</term>
<term>Approche probabiliste</term>
</keywords>
<keywords scheme="Wicri" type="topic" xml:lang="fr"><term>Méthode statistique</term>
</keywords>
</textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">In this paper we propose a novel approach for writer adaptation in a handwritten word-spotting task. The method exploits the fact that the semi-continuous hidden Markov model separates the word model parameters into (i) a codebook of shapes and (ii) a set of word-specific parameters. Our main contribution is to employ this property to derive writer-specific word models by statistically adapting an initial universal codebook to each document. This process is unsupervised and does not even require the appearance of the keyword(s) in the searched document. Experimental results show an increase in performance when this adaptation technique is applied. To the best of our knowledge, this is the first work dealing with adaptation for word-spotting. The preliminary version of this paper obtained an IBM Best Student Paper Award at the 19th International Conference on Pattern Recognition.</div>
</front>
</TEI>
<affiliations><list><country><li>Espagne</li>
<li>France</li>
</country>
<region><li>Auvergne-Rhône-Alpes</li>
<li>Catalogne</li>
<li>Rhône-Alpes</li>
</region>
<settlement><li>Barcelone</li>
<li>Meylan</li>
</settlement>
<orgName><li>Université autonome de Barcelone</li>
</orgName>
</list>
<tree><country name="France"><region name="Auvergne-Rhône-Alpes"><name sortKey="Rodriguez Serrano, Jose A" sort="Rodriguez Serrano, Jose A" uniqKey="Rodriguez Serrano J" first="José A." last="Rodriguez-Serrano">José A. Rodriguez-Serrano</name>
</region>
<name sortKey="Perronnin, Florent" sort="Perronnin, Florent" uniqKey="Perronnin F" first="Florent" last="Perronnin">Florent Perronnin</name>
</country>
<country name="Espagne"><region name="Catalogne"><name sortKey="Rodriguez Serrano, Jose A" sort="Rodriguez Serrano, Jose A" uniqKey="Rodriguez Serrano J" first="José A." last="Rodriguez-Serrano">José A. Rodriguez-Serrano</name>
</region>
<name sortKey="Llados, Josep" sort="Llados, Josep" uniqKey="Llados J" first="Josep" last="Llados">Josep Llados</name>
<name sortKey="Sanchez, Gemma" sort="Sanchez, Gemma" uniqKey="Sanchez G" first="Gemma" last="Sanchez">Gemma Sanchez</name>
</country>
</tree>
</affiliations>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Asie/explor/AustralieFrV1/Data/Main/Exploration
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 007426 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 007426 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Wicri/Asie |area= AustralieFrV1 |flux= Main |étape= Exploration |type= RBID |clé= Pascal:10-0223708 |texte= Unsupervised writer adaptation of whole-word HMMs with application to word-spotting }}
This area was generated with Dilib version V0.6.33. |