Dense Segmentation-Aware Descriptors
Dense descriptors are becoming increasingly popular in a host of tasks, such as dense image correspondence, bag-of-words image classification, and label transfer. However, the extraction of descriptors on generic image points, rather than selecting geometric features, requires rethinking how to achi...
Gespeichert in:
Hauptverfasser: | , , , |
---|---|
Format: | Buchkapitel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | 107 |
---|---|
container_issue | |
container_start_page | 83 |
container_title | |
container_volume | |
creator | Trulls, Eduard Kokkinos, Iasonas Sanfeliu, Alberto Moreno-Noguer, Francesc |
description | Dense descriptors are becoming increasingly popular in a host of tasks, such as dense image correspondence, bag-of-words image classification, and label transfer. However, the extraction of descriptors on generic image points, rather than selecting geometric features, requires rethinking how to achieve invariance to nuisance parameters. In this work we pursue invariance to occlusions and background changes by introducing segmentation information within dense feature construction. The core idea is to use the segmentation cues to downplay the features coming from image areas that are unlikely to belong to the same region as the feature point. We show how to integrate this idea with dense SIFT, as well as with the dense scale- and rotation-invariant descriptor (SID). We thereby deliver dense descriptors that are invariant to background changes, rotation, and/or scaling. We explore the merit of our technique in conjunction with large displacement motion estimation and wide-baseline stereo, and demonstrate that exploiting segmentation information yields clear improvements. |
doi_str_mv | 10.1007/978-3-319-23048-1_5 |
format | Book Chapter |
fullrecord | <record><control><sourceid>csuc_XX2</sourceid><recordid>TN_cdi_csuc_recercat_oai_recercat_cat_2072_260614</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>oai_recercat_cat_2072_260614</sourcerecordid><originalsourceid>FETCH-LOGICAL-c1545-32af1155574f0484bdb812ce0742fa06a5c3ad5ed9f723c9638968f50933787a3</originalsourceid><addsrcrecordid>eNpFkE1LAzEQhiMiqLW_wEsPXqOTTL72WFqtQsGDeh6y2aSsH7slWfHvu63CHoaXZ2DegYexawG3AsDeVdZx5CgqLhGU44L0CbvEcXFkdTqBNedsXso7AAjtlHbugt2sY1fi4iXuvmI3-KHtO7788Tku1rGE3O6HPpcrdpb8Z4nz_5yxt4f719Uj3z5vnlbLLQ9CK81R-iSE1tqqdHhdN7UTMkSwSiYPxuuAvtGxqZKVGCqDrjIuaagQrbMeZ0z89YbyHSjHEHPwA_W-neAwEqwkacAINd2UfW67XcxU9_1HIQF08EOjH0IaDdDRB41-8BfuJVW0</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>book_chapter</recordtype></control><display><type>book_chapter</type><title>Dense Segmentation-Aware Descriptors</title><source>Recercat</source><creator>Trulls, Eduard ; Kokkinos, Iasonas ; Sanfeliu, Alberto ; Moreno-Noguer, Francesc</creator><contributor>Liu, Ce ; Hassner, Tal</contributor><creatorcontrib>Trulls, Eduard ; Kokkinos, Iasonas ; Sanfeliu, Alberto ; Moreno-Noguer, Francesc ; Liu, Ce ; Hassner, Tal</creatorcontrib><description>Dense descriptors are becoming increasingly popular in a host of tasks, such as dense image correspondence, bag-of-words image classification, and label transfer. However, the extraction of descriptors on generic image points, rather than selecting geometric features, requires rethinking how to achieve invariance to nuisance parameters. In this work we pursue invariance to occlusions and background changes by introducing segmentation information within dense feature construction. The core idea is to use the segmentation cues to downplay the features coming from image areas that are unlikely to belong to the same region as the feature point. We show how to integrate this idea with dense SIFT, as well as with the dense scale- and rotation-invariant descriptor (SID). We thereby deliver dense descriptors that are invariant to background changes, rotation, and/or scaling. We explore the merit of our technique in conjunction with large displacement motion estimation and wide-baseline stereo, and demonstrate that exploiting segmentation information yields clear improvements.</description><identifier>ISBN: 3319230476</identifier><identifier>ISBN: 9783319230474</identifier><identifier>ISBN: 3319230484</identifier><identifier>ISBN: 9783319230481</identifier><identifier>EISBN: 3319230484</identifier><identifier>EISBN: 9783319230481</identifier><identifier>DOI: 10.1007/978-3-319-23048-1_5</identifier><language>eng</language><publisher>Cham: Springer International Publishing</publisher><subject>Background Change ; Computer vision ; feature extraction ; Frame Pair ; Informàtica ; Intel·ligència artificial ; Optical Flow ; Segmentation Mask ; Sift Descriptor ; Visió per ordinador ; Àrees temàtiques de la UPC</subject><ispartof>Dense Image Correspondences for Computer Vision, 2016, p.83-107</ispartof><rights>Springer International Publishing Switzerland 2016</rights><rights>info:eu-repo/semantics/openAccess</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>230,779,780,885,26974</link.rule.ids><linktorsrc>$$Uhttps://recercat.cat/handle/2072/260614$$EView_record_in_Consorci_de_Serveis_Universitaris_de_Catalunya_(CSUC)$$FView_record_in_$$GConsorci_de_Serveis_Universitaris_de_Catalunya_(CSUC)$$Hfree_for_read</linktorsrc></links><search><contributor>Liu, Ce</contributor><contributor>Hassner, Tal</contributor><creatorcontrib>Trulls, Eduard</creatorcontrib><creatorcontrib>Kokkinos, Iasonas</creatorcontrib><creatorcontrib>Sanfeliu, Alberto</creatorcontrib><creatorcontrib>Moreno-Noguer, Francesc</creatorcontrib><title>Dense Segmentation-Aware Descriptors</title><title>Dense Image Correspondences for Computer Vision</title><description>Dense descriptors are becoming increasingly popular in a host of tasks, such as dense image correspondence, bag-of-words image classification, and label transfer. However, the extraction of descriptors on generic image points, rather than selecting geometric features, requires rethinking how to achieve invariance to nuisance parameters. In this work we pursue invariance to occlusions and background changes by introducing segmentation information within dense feature construction. The core idea is to use the segmentation cues to downplay the features coming from image areas that are unlikely to belong to the same region as the feature point. We show how to integrate this idea with dense SIFT, as well as with the dense scale- and rotation-invariant descriptor (SID). We thereby deliver dense descriptors that are invariant to background changes, rotation, and/or scaling. We explore the merit of our technique in conjunction with large displacement motion estimation and wide-baseline stereo, and demonstrate that exploiting segmentation information yields clear improvements.</description><subject>Background Change</subject><subject>Computer vision</subject><subject>feature extraction</subject><subject>Frame Pair</subject><subject>Informàtica</subject><subject>Intel·ligència artificial</subject><subject>Optical Flow</subject><subject>Segmentation Mask</subject><subject>Sift Descriptor</subject><subject>Visió per ordinador</subject><subject>Àrees temàtiques de la UPC</subject><isbn>3319230476</isbn><isbn>9783319230474</isbn><isbn>3319230484</isbn><isbn>9783319230481</isbn><isbn>3319230484</isbn><isbn>9783319230481</isbn><fulltext>true</fulltext><rsrctype>book_chapter</rsrctype><creationdate>2016</creationdate><recordtype>book_chapter</recordtype><sourceid>XX2</sourceid><recordid>eNpFkE1LAzEQhiMiqLW_wEsPXqOTTL72WFqtQsGDeh6y2aSsH7slWfHvu63CHoaXZ2DegYexawG3AsDeVdZx5CgqLhGU44L0CbvEcXFkdTqBNedsXso7AAjtlHbugt2sY1fi4iXuvmI3-KHtO7788Tku1rGE3O6HPpcrdpb8Z4nz_5yxt4f719Uj3z5vnlbLLQ9CK81R-iSE1tqqdHhdN7UTMkSwSiYPxuuAvtGxqZKVGCqDrjIuaagQrbMeZ0z89YbyHSjHEHPwA_W-neAwEqwkacAINd2UfW67XcxU9_1HIQF08EOjH0IaDdDRB41-8BfuJVW0</recordid><startdate>2016</startdate><enddate>2016</enddate><creator>Trulls, Eduard</creator><creator>Kokkinos, Iasonas</creator><creator>Sanfeliu, Alberto</creator><creator>Moreno-Noguer, Francesc</creator><general>Springer International Publishing</general><general>Springer</general><scope>XX2</scope></search><sort><creationdate>2016</creationdate><title>Dense Segmentation-Aware Descriptors</title><author>Trulls, Eduard ; Kokkinos, Iasonas ; Sanfeliu, Alberto ; Moreno-Noguer, Francesc</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c1545-32af1155574f0484bdb812ce0742fa06a5c3ad5ed9f723c9638968f50933787a3</frbrgroupid><rsrctype>book_chapters</rsrctype><prefilter>book_chapters</prefilter><language>eng</language><creationdate>2016</creationdate><topic>Background Change</topic><topic>Computer vision</topic><topic>feature extraction</topic><topic>Frame Pair</topic><topic>Informàtica</topic><topic>Intel·ligència artificial</topic><topic>Optical Flow</topic><topic>Segmentation Mask</topic><topic>Sift Descriptor</topic><topic>Visió per ordinador</topic><topic>Àrees temàtiques de la UPC</topic><toplevel>online_resources</toplevel><creatorcontrib>Trulls, Eduard</creatorcontrib><creatorcontrib>Kokkinos, Iasonas</creatorcontrib><creatorcontrib>Sanfeliu, Alberto</creatorcontrib><creatorcontrib>Moreno-Noguer, Francesc</creatorcontrib><collection>Recercat</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Trulls, Eduard</au><au>Kokkinos, Iasonas</au><au>Sanfeliu, Alberto</au><au>Moreno-Noguer, Francesc</au><au>Liu, Ce</au><au>Hassner, Tal</au><format>book</format><genre>bookitem</genre><ristype>CHAP</ristype><atitle>Dense Segmentation-Aware Descriptors</atitle><btitle>Dense Image Correspondences for Computer Vision</btitle><date>2016</date><risdate>2016</risdate><spage>83</spage><epage>107</epage><pages>83-107</pages><isbn>3319230476</isbn><isbn>9783319230474</isbn><isbn>3319230484</isbn><isbn>9783319230481</isbn><eisbn>3319230484</eisbn><eisbn>9783319230481</eisbn><abstract>Dense descriptors are becoming increasingly popular in a host of tasks, such as dense image correspondence, bag-of-words image classification, and label transfer. However, the extraction of descriptors on generic image points, rather than selecting geometric features, requires rethinking how to achieve invariance to nuisance parameters. In this work we pursue invariance to occlusions and background changes by introducing segmentation information within dense feature construction. The core idea is to use the segmentation cues to downplay the features coming from image areas that are unlikely to belong to the same region as the feature point. We show how to integrate this idea with dense SIFT, as well as with the dense scale- and rotation-invariant descriptor (SID). We thereby deliver dense descriptors that are invariant to background changes, rotation, and/or scaling. We explore the merit of our technique in conjunction with large displacement motion estimation and wide-baseline stereo, and demonstrate that exploiting segmentation information yields clear improvements.</abstract><cop>Cham</cop><pub>Springer International Publishing</pub><doi>10.1007/978-3-319-23048-1_5</doi><tpages>25</tpages><oa>free_for_read</oa></addata></record> |
fulltext | fulltext_linktorsrc |
identifier | ISBN: 3319230476 |
ispartof | Dense Image Correspondences for Computer Vision, 2016, p.83-107 |
issn | |
language | eng |
recordid | cdi_csuc_recercat_oai_recercat_cat_2072_260614 |
source | Recercat |
subjects | Background Change Computer vision feature extraction Frame Pair Informàtica Intel·ligència artificial Optical Flow Segmentation Mask Sift Descriptor Visió per ordinador Àrees temàtiques de la UPC |
title | Dense Segmentation-Aware Descriptors |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T15%3A45%3A42IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-csuc_XX2&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=bookitem&rft.atitle=Dense%20Segmentation-Aware%20Descriptors&rft.btitle=Dense%20Image%20Correspondences%20for%20Computer%20Vision&rft.au=Trulls,%20Eduard&rft.date=2016&rft.spage=83&rft.epage=107&rft.pages=83-107&rft.isbn=3319230476&rft.isbn_list=9783319230474&rft.isbn_list=3319230484&rft.isbn_list=9783319230481&rft_id=info:doi/10.1007/978-3-319-23048-1_5&rft_dat=%3Ccsuc_XX2%3Eoai_recercat_cat_2072_260614%3C/csuc_XX2%3E%3Curl%3E%3C/url%3E&rft.eisbn=3319230484&rft.eisbn_list=9783319230481&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |