OBJECT RETRIEVAL SYSTEM, OBJECT RETRIEVAL METHOD AND LEARNED MODEL

To provide a technology capable of specifying a target object even in a situation where there are a plurality of candidates of the target object.SOLUTION: An object retrieval system includes: an instruction statement acquisition unit that acquires an instruction statement relating to a specific obje...

Full description

Saved in:
Bibliographic Details
Main Authors SUGIURA KOMEI, KAWAI HISASHI, MAGASSOUBA ALY
Format Patent
LanguageEnglish
Japanese
Published 26.11.2020
Subjects
Online AccessGet full text

Cover

Loading…
Abstract To provide a technology capable of specifying a target object even in a situation where there are a plurality of candidates of the target object.SOLUTION: An object retrieval system includes: an instruction statement acquisition unit that acquires an instruction statement relating to a specific object; an image extraction unit that extracts one or a plurality of first partial images indicating an individual object included in an input image from the input image associated with the instruction statement; and a learned model that receives an input of the instruction statement, each of the first partial images, and information indicating an intra-image environment of the first partial image to output a probability that each of the first partial images is an object specified by the instruction statement. The learned model is learned by the instruction statement specifying any object included in the image, and the partial image indicating the object specified by the instruction statement.SELECTED DRAWING: Figure 1 【課題】対象となる物体の候補が複数存在するような状況であっても、対象となる物体を特定できる技術を提供する。【解決手段】対象物検索システムは、特定の対象物に関する命令文を取得する命令文取得部と、命令文に関連付けられた入力画像から、当該入力画像に含まれる個々の物体を示す1または複数の第1の部分画像を抽出する画像抽出部と、命令文と、第1の部分画像の各々と、当該第1の部分画像の画像内環境を示す情報との入力を受けて、第1の部分画像の各々が命令文により特定される対象物である確率を出力する学習済モデルとを含む。学習済モデルは、画像に含まれるいずれかの物体を特定する命令文と、当該命令文により特定される物体を示す部分画像とを含むトレーニングデータセットにより学習されている。【選択図】図1
AbstractList To provide a technology capable of specifying a target object even in a situation where there are a plurality of candidates of the target object.SOLUTION: An object retrieval system includes: an instruction statement acquisition unit that acquires an instruction statement relating to a specific object; an image extraction unit that extracts one or a plurality of first partial images indicating an individual object included in an input image from the input image associated with the instruction statement; and a learned model that receives an input of the instruction statement, each of the first partial images, and information indicating an intra-image environment of the first partial image to output a probability that each of the first partial images is an object specified by the instruction statement. The learned model is learned by the instruction statement specifying any object included in the image, and the partial image indicating the object specified by the instruction statement.SELECTED DRAWING: Figure 1 【課題】対象となる物体の候補が複数存在するような状況であっても、対象となる物体を特定できる技術を提供する。【解決手段】対象物検索システムは、特定の対象物に関する命令文を取得する命令文取得部と、命令文に関連付けられた入力画像から、当該入力画像に含まれる個々の物体を示す1または複数の第1の部分画像を抽出する画像抽出部と、命令文と、第1の部分画像の各々と、当該第1の部分画像の画像内環境を示す情報との入力を受けて、第1の部分画像の各々が命令文により特定される対象物である確率を出力する学習済モデルとを含む。学習済モデルは、画像に含まれるいずれかの物体を特定する命令文と、当該命令文により特定される物体を示す部分画像とを含むトレーニングデータセットにより学習されている。【選択図】図1
Author MAGASSOUBA ALY
KAWAI HISASHI
SUGIURA KOMEI
Author_xml – fullname: SUGIURA KOMEI
– fullname: KAWAI HISASHI
– fullname: MAGASSOUBA ALY
BookMark eNrjYmDJy89L5WRw8nfycnUOUQhyDQnydA1z9FEIjgwOcfXVUcCQ8HUN8fB3UXD0c1HwcXUM8nN1UfD1d3H14WFgTUvMKU7lhdLcDEpuriHOHrqpBfnxqcUFicmpeakl8V4BRgZGBoaWBpbGBo7GRCkCAF_dLSo
ContentType Patent
DBID EVB
DatabaseName esp@cenet
DatabaseTitleList
Database_xml – sequence: 1
  dbid: EVB
  name: esp@cenet
  url: http://worldwide.espacenet.com/singleLineSearch?locale=en_EP
  sourceTypes: Open Access Repository
DeliveryMethod fulltext_linktorsrc
Discipline Medicine
Chemistry
Sciences
Physics
DocumentTitleAlternate 対象物検索システム、対象物検索方法および学習済モデル
ExternalDocumentID JP2020190930A
GroupedDBID EVB
ID FETCH-epo_espacenet_JP2020190930A3
IEDL.DBID EVB
IngestDate Fri Jul 19 15:12:00 EDT 2024
IsOpenAccess true
IsPeerReviewed false
IsScholarly false
Language English
Japanese
LinkModel DirectLink
MergedId FETCHMERGED-epo_espacenet_JP2020190930A3
Notes Application Number: JP20190095922
OpenAccessLink https://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20201126&DB=EPODOC&CC=JP&NR=2020190930A
ParticipantIDs epo_espacenet_JP2020190930A
PublicationCentury 2000
PublicationDate 20201126
PublicationDateYYYYMMDD 2020-11-26
PublicationDate_xml – month: 11
  year: 2020
  text: 20201126
  day: 26
PublicationDecade 2020
PublicationYear 2020
RelatedCompanies NATIONAL INSTITUTE OF INFORMATION & COMMUNICATIONS TECHNOLOGY
RelatedCompanies_xml – name: NATIONAL INSTITUTE OF INFORMATION & COMMUNICATIONS TECHNOLOGY
Score 3.4260137
Snippet To provide a technology capable of specifying a target object even in a situation where there are a plurality of candidates of the target object.SOLUTION: An...
SourceID epo
SourceType Open Access Repository
SubjectTerms CALCULATING
COMPUTING
COUNTING
ELECTRIC DIGITAL DATA PROCESSING
IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
PHYSICS
Title OBJECT RETRIEVAL SYSTEM, OBJECT RETRIEVAL METHOD AND LEARNED MODEL
URI https://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20201126&DB=EPODOC&locale=&CC=JP&NR=2020190930A
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwfV1LS8NAEB5qfd60KmpVFpGcLDaPJs2hSJLdEEPzII2lnkoTE1ChLTbi33d2bbUg9LgzMOwOzM43u_MAuJXliVnKPOepMNstrZtrLcTNPBNARXCayYjQ-Y9uEOrek-aPOqMavK9qYUSf0C_RHBEtKkd7r8R9Pf97xKIit3Jxn70iafbgpj0qLaNj7s0UXaJ2j8URjRzJcXp-LIXJD8_E8L1tbcE24miDmwMb2rwsZb7uU9xD2IlR3LQ6gtrbpAH7zmr0WgP2guWPdwN2RYpmvkDi0gwXx2BHYkoJSViaPLKh1SeD50HKgjvyjxGw1IsosUJK-sxKQkZJEFHWP4Ebl6WO18JtjX-VMPbjtSOop1CfzqbFGRC1NIqJkZWGkXe00jRQ86Uu5_mL0kWsp5jn0Nwg6GIjtwkHfMVL7xT9EurVx2dxhT64yq6F7r4BXqWCPw
link.rule.ids 230,309,783,888,25576,76876
linkProvider European Patent Office
linkToHtml http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwfV1LS8NAEB5qfdSbVkWtj0UkJ4tNmibNoUia3ZLWvEhjqaeSxARUaIuN-PedXVstCL3OwLA7MDvf7LwAbmU5NnKZ1zxlRqOutlO1jriZVwI0EZwmMiJ0ntF1Pc1-Ugfj1rgE76teGDEn9EsMR0SLStHeC_Fez_8-saiorVzcJ69Imj30og6VltEx92aKJtFuhwU-9S3JsjqDQPLCH56B4XvD3IJtxNg6Nwc26vK2lPm6T-kdwE6A4qbFIZTe4ipUrNXqtSrsucuMdxV2RYlmukDi0gwXR9D1xZYSErIo7LOR6ZDh8zBi7h35x3BZZPuUmB4lDjNDj1Hi-pQ5x3DTY5Fl1_FYk18lTAbB2hWaJ1CezqbZKZBmrmexnuS6nrbU3NBR87kmp-mL0kaspxhnUNsg6Hwj9xoqduQ6E6fvPdZgn3N4G56iXUC5-PjMLtEfF8mV0OM3prWFMg
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Apatent&rft.title=OBJECT+RETRIEVAL+SYSTEM%2C+OBJECT+RETRIEVAL+METHOD+AND+LEARNED+MODEL&rft.inventor=SUGIURA+KOMEI&rft.inventor=KAWAI+HISASHI&rft.inventor=MAGASSOUBA+ALY&rft.date=2020-11-26&rft.externalDBID=A&rft.externalDocID=JP2020190930A