Large scale document image retrieval by automatic word annotation
Identifieur interne : 000128 ( Main/Exploration ); précédent : 000127; suivant : 000129Large scale document image retrieval by automatic word annotation
Auteurs : K. Pramod Sankar [Inde] ; R. Manmatha [États-Unis] ; C. V. Jawahar [Inde]Source :
- International journal on document analysis and recognition : (Print) [ 1433-2833 ] ; 2014.
Descripteurs français
- Pascal (Inist)
- Wicri :
- topic : Classification.
English descriptors
- KwdEn :
Abstract
In this paper, we present a practical and scalable retrieval framework for large-scale document image collections, for an Indian language script that does not have a robust OCR. OCR-based methods face difficulties in character segmentation and recognition, especially for the complex Indian language scripts. We realize that character recognition is only an intermediate step toward actually labeling words. Hence, we re-pose the problem as one of directly performing word annotation. This new approach has better recognition performance, as well as easier segmentation requirements. However, the number of classes in word annotation is much larger than those for character recognition, making such a classification scheme expensive to train and test. To address this issue, we present a novel framework that replaces naive classification with a carefully designed mixture of indexing and classification schemes. This enables us to build a search system over a large collection of 1,000 books of Telugu, consisting of 120K document images or 36M individual words. This is the largest searchable document image collection for a script without an OCR that we are aware of. Our retrieval system performs significantly well with a mean average precision of 0.8.
Affiliations:
Links toward previous steps (curation, corpus...)
- to stream PascalFrancis, to step Corpus: 000011
- to stream PascalFrancis, to step Curation: 000754
- to stream PascalFrancis, to step Checkpoint: 000011
- to stream Main, to step Merge: 000129
- to stream Main, to step Curation: 000128
Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="en" level="a">Large scale document image retrieval by automatic word annotation</title><author><name sortKey="Pramod Sankar, K" sort="Pramod Sankar, K" uniqKey="Pramod Sankar K" first="K." last="Pramod Sankar">K. Pramod Sankar</name><affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Xerox Research Center India</s1><s2>Bengaluru</s2><s3>IND</s3><sZ>1 aut.</sZ></inist:fA14><country>Inde</country><wicri:noRegion>Xerox Research Center India</wicri:noRegion></affiliation></author><author><name sortKey="Manmatha, R" sort="Manmatha, R" uniqKey="Manmatha R" first="R." last="Manmatha">R. Manmatha</name><affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Multi-media Indexing and Retrieval Group, Department of Computer Science, University of Massachusetts</s1><s2>Amherst, MA</s2><s3>USA</s3><sZ>2 aut.</sZ></inist:fA14><country>États-Unis</country><placeName><region type="state">Massachusetts</region><settlement type="city">Amherst (Massachusetts)</settlement></placeName><orgName type="university">Université du Massachusetts</orgName></affiliation></author><author><name sortKey="Jawahar, C V" sort="Jawahar, C V" uniqKey="Jawahar C" first="C. V." last="Jawahar">C. V. Jawahar</name><affiliation wicri:level="1"><inist:fA14 i1="03"><s1>Center for Visual Information Technology, IIIT-Hyderabad</s1><s2>Hyderabad</s2><s3>IND</s3><sZ>3 aut.</sZ></inist:fA14><country>Inde</country><wicri:noRegion>Hyderabad</wicri:noRegion></affiliation></author></titleStmt><publicationStmt><idno type="wicri:source">INIST</idno><idno type="inist">14-0199538</idno><date when="2014">2014</date><idno type="stanalyst">PASCAL 14-0199538 INIST</idno><idno type="RBID">Pascal:14-0199538</idno><idno type="wicri:Area/PascalFrancis/Corpus">000011</idno><idno type="wicri:Area/PascalFrancis/Curation">000754</idno><idno type="wicri:Area/PascalFrancis/Checkpoint">000011</idno><idno type="wicri:doubleKey">1433-2833:2014:Pramod Sankar K:large:scale:document</idno><idno type="wicri:Area/Main/Merge">000129</idno><idno type="wicri:Area/Main/Curation">000128</idno><idno type="wicri:Area/Main/Exploration">000128</idno></publicationStmt><sourceDesc><biblStruct><analytic><title xml:lang="en" level="a">Large scale document image retrieval by automatic word annotation</title><author><name sortKey="Pramod Sankar, K" sort="Pramod Sankar, K" uniqKey="Pramod Sankar K" first="K." last="Pramod Sankar">K. Pramod Sankar</name><affiliation wicri:level="1"><inist:fA14 i1="01"><s1>Xerox Research Center India</s1><s2>Bengaluru</s2><s3>IND</s3><sZ>1 aut.</sZ></inist:fA14><country>Inde</country><wicri:noRegion>Xerox Research Center India</wicri:noRegion></affiliation></author><author><name sortKey="Manmatha, R" sort="Manmatha, R" uniqKey="Manmatha R" first="R." last="Manmatha">R. Manmatha</name><affiliation wicri:level="4"><inist:fA14 i1="02"><s1>Multi-media Indexing and Retrieval Group, Department of Computer Science, University of Massachusetts</s1><s2>Amherst, MA</s2><s3>USA</s3><sZ>2 aut.</sZ></inist:fA14><country>États-Unis</country><placeName><region type="state">Massachusetts</region><settlement type="city">Amherst (Massachusetts)</settlement></placeName><orgName type="university">Université du Massachusetts</orgName></affiliation></author><author><name sortKey="Jawahar, C V" sort="Jawahar, C V" uniqKey="Jawahar C" first="C. V." last="Jawahar">C. V. Jawahar</name><affiliation wicri:level="1"><inist:fA14 i1="03"><s1>Center for Visual Information Technology, IIIT-Hyderabad</s1><s2>Hyderabad</s2><s3>IND</s3><sZ>3 aut.</sZ></inist:fA14><country>Inde</country><wicri:noRegion>Hyderabad</wicri:noRegion></affiliation></author></analytic><series><title level="j" type="main">International journal on document analysis and recognition : (Print)</title><title level="j" type="abbreviated">Int. j. doc. anal. recognit. : (Print)</title><idno type="ISSN">1433-2833</idno><imprint><date when="2014">2014</date></imprint></series></biblStruct></sourceDesc><seriesStmt><title level="j" type="main">International journal on document analysis and recognition : (Print)</title><title level="j" type="abbreviated">Int. j. doc. anal. recognit. : (Print)</title><idno type="ISSN">1433-2833</idno></seriesStmt></fileDesc><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Annotation</term><term>Book</term><term>Character recognition</term><term>Classification</term><term>Classification scheme</term><term>Facies</term><term>Image databank</term><term>Image processing</term><term>Image retrieval</term><term>Image segmentation</term><term>Indexing</term><term>Information retrieval</term><term>Large scale</term><term>Optical character recognition</term><term>Pattern recognition</term><term>Very large databases</term></keywords><keywords scheme="Pascal" xml:lang="fr"><term>Recherche information</term><term>Banque image</term><term>Reconnaissance caractère</term><term>Reconnaissance optique caractère</term><term>Reconnaissance forme</term><term>Traitement image</term><term>Classification</term><term>Indexation</term><term>Annotation</term><term>Faciès</term><term>Plan classement</term><term>Livre</term><term>Echelle grande</term><term>.</term><term>Base donnée très grande</term><term>Segmentation image</term><term>Recherche image</term></keywords><keywords scheme="Wicri" type="topic" xml:lang="fr"><term>Classification</term></keywords></textClass></profileDesc></teiHeader><front><div type="abstract" xml:lang="en">In this paper, we present a practical and scalable retrieval framework for large-scale document image collections, for an Indian language script that does not have a robust OCR. OCR-based methods face difficulties in character segmentation and recognition, especially for the complex Indian language scripts. We realize that character recognition is only an intermediate step toward actually labeling words. Hence, we re-pose the problem as one of directly performing word annotation. This new approach has better recognition performance, as well as easier segmentation requirements. However, the number of classes in word annotation is much larger than those for character recognition, making such a classification scheme expensive to train and test. To address this issue, we present a novel framework that replaces naive classification with a carefully designed mixture of indexing and classification schemes. This enables us to build a search system over a large collection of 1,000 books of Telugu, consisting of 120K document images or 36M individual words. This is the largest searchable document image collection for a script without an OCR that we are aware of. Our retrieval system performs significantly well with a mean average precision of 0.8.</div></front></TEI><affiliations><list><country><li>Inde</li><li>États-Unis</li></country><region><li>Massachusetts</li></region><settlement><li>Amherst (Massachusetts)</li></settlement><orgName><li>Université du Massachusetts</li></orgName></list><tree><country name="Inde"><noRegion><name sortKey="Pramod Sankar, K" sort="Pramod Sankar, K" uniqKey="Pramod Sankar K" first="K." last="Pramod Sankar">K. Pramod Sankar</name></noRegion><name sortKey="Jawahar, C V" sort="Jawahar, C V" uniqKey="Jawahar C" first="C. V." last="Jawahar">C. V. Jawahar</name></country><country name="États-Unis"><region name="Massachusetts"><name sortKey="Manmatha, R" sort="Manmatha, R" uniqKey="Manmatha R" first="R." last="Manmatha">R. Manmatha</name></region></country></tree></affiliations></record>Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Ticri/CIDE/explor/OcrV1/Data/Main/Exploration
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 000128 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 000128 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Ticri/CIDE
|area= OcrV1
|flux= Main
|étape= Exploration
|type= RBID
|clé= Pascal:14-0199538
|texte= Large scale document image retrieval by automatic word annotation
}}
| This area was generated with Dilib version V0.6.32. |  |