What to Learn: Features, Image Transformations, or Both?

Long-term visual localization is an essential problem in robotics and computer vision, but remains challenging due to the environmental appearance changes caused by lighting and seasons. While many existing works have attempted to solve it by directly learning invariant sparse keypoints and descript...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2023-06
Hauptverfasser: Chen, Yuxuan, Xu, Binbin, Dümbgen, Frederike, Barfoot, Timothy D
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Chen, Yuxuan
Xu, Binbin
Dümbgen, Frederike
Barfoot, Timothy D
description Long-term visual localization is an essential problem in robotics and computer vision, but remains challenging due to the environmental appearance changes caused by lighting and seasons. While many existing works have attempted to solve it by directly learning invariant sparse keypoints and descriptors to match scenes, these approaches still struggle with adverse appearance changes. Recent developments in image transformations such as neural style transfer have emerged as an alternative to address such appearance gaps. In this work, we propose to combine an image transformation network and a feature-learning network to improve long-term localization performance. Given night-to-day image pairs, the image transformation network transforms the night images into day-like conditions prior to feature matching; the feature network learns to detect keypoint locations with their associated descriptor values, which can be passed to a classical pose estimator to compute the relative poses. We conducted various experiments to examine the effectiveness of combining style transfer and feature learning and its training strategy, showing that such a combination greatly improves long-term localization performance.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2828968633</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2828968633</sourcerecordid><originalsourceid>FETCH-proquest_journals_28289686333</originalsourceid><addsrcrecordid>eNqNikEKwjAQAIMgWLR_CHi1EDc2Ri-CYlHwWPBY9pBai83qJvm_PfgATwMzMxEZaL0u7AZgJvIQeqUUmC2Upc6EvXcYZSR5c8h-LyuHMbELK3kd8OFkzehDSzxgfJIfNbE8UuwOCzFt8RVc_uNcLKtzfboUb6ZPciE2PSX2Y2rAgt0Za7TW_11fxWc1Hw</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2828968633</pqid></control><display><type>article</type><title>What to Learn: Features, Image Transformations, or Both?</title><source>Freely Accessible Journals</source><creator>Chen, Yuxuan ; Xu, Binbin ; Dümbgen, Frederike ; Barfoot, Timothy D</creator><creatorcontrib>Chen, Yuxuan ; Xu, Binbin ; Dümbgen, Frederike ; Barfoot, Timothy D</creatorcontrib><description>Long-term visual localization is an essential problem in robotics and computer vision, but remains challenging due to the environmental appearance changes caused by lighting and seasons. While many existing works have attempted to solve it by directly learning invariant sparse keypoints and descriptors to match scenes, these approaches still struggle with adverse appearance changes. Recent developments in image transformations such as neural style transfer have emerged as an alternative to address such appearance gaps. In this work, we propose to combine an image transformation network and a feature-learning network to improve long-term localization performance. Given night-to-day image pairs, the image transformation network transforms the night images into day-like conditions prior to feature matching; the feature network learns to detect keypoint locations with their associated descriptor values, which can be passed to a classical pose estimator to compute the relative poses. We conducted various experiments to examine the effectiveness of combining style transfer and feature learning and its training strategy, showing that such a combination greatly improves long-term localization performance.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Computer vision ; Localization ; Night ; Robotics ; Transformations</subject><ispartof>arXiv.org, 2023-06</ispartof><rights>2023. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</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>781,785</link.rule.ids></links><search><creatorcontrib>Chen, Yuxuan</creatorcontrib><creatorcontrib>Xu, Binbin</creatorcontrib><creatorcontrib>Dümbgen, Frederike</creatorcontrib><creatorcontrib>Barfoot, Timothy D</creatorcontrib><title>What to Learn: Features, Image Transformations, or Both?</title><title>arXiv.org</title><description>Long-term visual localization is an essential problem in robotics and computer vision, but remains challenging due to the environmental appearance changes caused by lighting and seasons. While many existing works have attempted to solve it by directly learning invariant sparse keypoints and descriptors to match scenes, these approaches still struggle with adverse appearance changes. Recent developments in image transformations such as neural style transfer have emerged as an alternative to address such appearance gaps. In this work, we propose to combine an image transformation network and a feature-learning network to improve long-term localization performance. Given night-to-day image pairs, the image transformation network transforms the night images into day-like conditions prior to feature matching; the feature network learns to detect keypoint locations with their associated descriptor values, which can be passed to a classical pose estimator to compute the relative poses. We conducted various experiments to examine the effectiveness of combining style transfer and feature learning and its training strategy, showing that such a combination greatly improves long-term localization performance.</description><subject>Computer vision</subject><subject>Localization</subject><subject>Night</subject><subject>Robotics</subject><subject>Transformations</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNqNikEKwjAQAIMgWLR_CHi1EDc2Ri-CYlHwWPBY9pBai83qJvm_PfgATwMzMxEZaL0u7AZgJvIQeqUUmC2Upc6EvXcYZSR5c8h-LyuHMbELK3kd8OFkzehDSzxgfJIfNbE8UuwOCzFt8RVc_uNcLKtzfboUb6ZPciE2PSX2Y2rAgt0Za7TW_11fxWc1Hw</recordid><startdate>20230622</startdate><enddate>20230622</enddate><creator>Chen, Yuxuan</creator><creator>Xu, Binbin</creator><creator>Dümbgen, Frederike</creator><creator>Barfoot, Timothy D</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20230622</creationdate><title>What to Learn: Features, Image Transformations, or Both?</title><author>Chen, Yuxuan ; Xu, Binbin ; Dümbgen, Frederike ; Barfoot, Timothy D</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_28289686333</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer vision</topic><topic>Localization</topic><topic>Night</topic><topic>Robotics</topic><topic>Transformations</topic><toplevel>online_resources</toplevel><creatorcontrib>Chen, Yuxuan</creatorcontrib><creatorcontrib>Xu, Binbin</creatorcontrib><creatorcontrib>Dümbgen, Frederike</creatorcontrib><creatorcontrib>Barfoot, Timothy D</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Access via ProQuest (Open Access)</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Chen, Yuxuan</au><au>Xu, Binbin</au><au>Dümbgen, Frederike</au><au>Barfoot, Timothy D</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>What to Learn: Features, Image Transformations, or Both?</atitle><jtitle>arXiv.org</jtitle><date>2023-06-22</date><risdate>2023</risdate><eissn>2331-8422</eissn><abstract>Long-term visual localization is an essential problem in robotics and computer vision, but remains challenging due to the environmental appearance changes caused by lighting and seasons. While many existing works have attempted to solve it by directly learning invariant sparse keypoints and descriptors to match scenes, these approaches still struggle with adverse appearance changes. Recent developments in image transformations such as neural style transfer have emerged as an alternative to address such appearance gaps. In this work, we propose to combine an image transformation network and a feature-learning network to improve long-term localization performance. Given night-to-day image pairs, the image transformation network transforms the night images into day-like conditions prior to feature matching; the feature network learns to detect keypoint locations with their associated descriptor values, which can be passed to a classical pose estimator to compute the relative poses. We conducted various experiments to examine the effectiveness of combining style transfer and feature learning and its training strategy, showing that such a combination greatly improves long-term localization performance.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2023-06
issn 2331-8422
language eng
recordid cdi_proquest_journals_2828968633
source Freely Accessible Journals
subjects Computer vision
Localization
Night
Robotics
Transformations
title What to Learn: Features, Image Transformations, or Both?
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-16T06%3A50%3A23IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=What%20to%20Learn:%20Features,%20Image%20Transformations,%20or%20Both?&rft.jtitle=arXiv.org&rft.au=Chen,%20Yuxuan&rft.date=2023-06-22&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2828968633%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2828968633&rft_id=info:pmid/&rfr_iscdi=true