Anvil a system for the retrieval of captioned images using nlp techniques

Anvil a system for the retrieval of captioned images using nlp techniques

ID:37657487

大小:245.18 KB

页数:13页

时间:2019-05-27

Anvil a system for the retrieval of captioned images using nlp techniques_第1页
Anvil a system for the retrieval of captioned images using nlp techniques_第2页
Anvil a system for the retrieval of captioned images using nlp techniques_第3页
Anvil a system for the retrieval of captioned images using nlp techniques_第4页
Anvil a system for the retrieval of captioned images using nlp techniques_第5页
资源描述:

《Anvil a system for the retrieval of captioned images using nlp techniques》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库

1、ANVIL:aSystemfortheRetrievalofCaptionedImagesusingNLPTechniquesTonyRose,DavidElworthy,AaronKotcheffandAmandaClareCanonResearchCentreEuropeGuildford,SurreyPetrosTsonisDept.ofComputingScience,UniversityofGlasgowScotlandAbstractANVILisasystemdesignedfortheretrievalofimagesannotatedwith

2、shortcaptions.ItusesNLPtechniquestoextractdependencystructuresfromcaptionsandqueries,andthenappliesarobustmatchingalgorithmtorecursivelyexploreandcomparethem.TherearecurrentlytwomaininterfacestoANVIL:alist-baseddisplayanda2Dspatiallayoutthatallowsuserstointeractwithandnavigatebetwee

3、nsimilarimages.ANVILwasdesignedtooperateaspartofapubliclyaccessible,WWW-basedimageretrievalserver.Consequently,product-levelengineeringstandardswererequired.Thispaperexaminesboththeresearchaspectsofthesystemandalsolooksatsomeofthedesignandevaluationissues.1IntroductionAfundamentalai

4、mofmanyInformationRetrieval(IR)systemsistomatchauser’sstatementoftheirinformationneedwiththecontentsofadocumentcollection.Sincequeriesanddocumentsareoftenpredominantlytextual,itseemsreasonabletosuggestthatinterpretationoftheirunderlyinglinguisticstructurewouldbebeneficial.However,in

5、practice,thishasrarelyprovedtobethecase[1].OnenotableexceptionistheIntermezzosystem[2],inwhichNLPtechniqueswereappliedtotheretrievalofimagesthathadbeenannotatedwithshortcaptions.Thissystemreturnedaprecisionofaround90%for50queriesona500,000-imagedatabase.Thisresultwouldtendtoindicate

6、thatthecombinationofashortcaption(asopposedtoalargerdocument)withsomecontroloveritscontent(asopposedtominimaleditorialcontroloverlargedocumentcollections)mayfacilitatetheapplicationofspecificNLPtechniquestoimprovetheretrievalprocess.Thispaper,andtheANVILimageretrievalsystemitdescrib

7、es,constitutesafurtherinvestigationofthishypothesis.1.1DesignGoalsANVILisdesignedfortheretrievalofcaptionedimages,usingfastandaccuratenaturallanguagetechniques.AswithIntermezzo,weuseadatabaseofimageswithphrasalcaptions,theirlengthrangingfrom1to20words(mean=~9words).Someexamplesareas

8、follows:Goldencolum

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。