EvRepSL: Event-Stream Representation via Self-Supervised Learning for Event-Based Vision
Event-stream representation is the first step for many computer vision tasks using event cameras. It converts the asynchronous event-streams into a formatted structure so that conventional machine learning models can be applied easily. However, most of the state-of-the-art event-stream representatio...
Gespeichert in:
Veröffentlicht in: | IEEE transactions on image processing 2024, Vol.33, p.6579-6591 |
---|---|
Hauptverfasser: | , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | 6591 |
---|---|
container_issue | |
container_start_page | 6579 |
container_title | IEEE transactions on image processing |
container_volume | 33 |
creator | Qu, Qiang Chen, Xiaoming Ying Chung, Yuk Shen, Yiran |
description | Event-stream representation is the first step for many computer vision tasks using event cameras. It converts the asynchronous event-streams into a formatted structure so that conventional machine learning models can be applied easily. However, most of the state-of-the-art event-stream representations are manually designed and the quality of these representations cannot be guaranteed due to the noisy nature of event-streams. In this paper, we introduce a data-driven approach aiming at enhancing the quality of event-stream representations. Our approach commences with the introduction of a new event-stream representation based on spatial-temporal statistics, denoted as EvRep. Subsequently, we theoretically derive the intrinsic relationship between asynchronous event-streams and synchronous video frames. Building upon this theoretical relationship, we train a representation generator, RepGen, in a self-supervised learning manner accepting EvRep as input. Finally, the event-streams are converted to high-quality representations, termed as EvRepSL, by going through the learned RepGen (without the need of fine-tuning or retraining). Our methodology is rigorously validated through extensive evaluations on a variety of mainstream event-based classification and optical flow datasets (captured with various types of event cameras). The experimental results highlight not only our approach's superior performance over existing event-stream representations but also its versatility, being agnostic to different event cameras and tasks. |
doi_str_mv | 10.1109/TIP.2024.3497795 |
format | Article |
fullrecord | <record><control><sourceid>proquest_RIE</sourceid><recordid>TN_cdi_proquest_journals_3131908816</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>10758409</ieee_id><sourcerecordid>3131908816</sourcerecordid><originalsourceid>FETCH-LOGICAL-c175t-57f163d3ed607aea94d1bb3c019598f61e73cb1d573bda0aa618d8fb2bfe0b9c3</originalsourceid><addsrcrecordid>eNpNkMFLwzAUxoMoOKd3Dx4KnjvzmqRpvOmYOigodoq3kLYv0rG1NekK_vdmbAdP7_G973sf_Ai5BjoDoOputXybJTThM8aVlEqckAkoDjGlPDkNOxUylsDVObnwfk0pcAHphHwtxnfsi_w-WozYDnExODTbKGgOfRDM0HRtNDYmKnBj42LXoxsbj3WUo3Ft035HtnPH8KPZHz4bHzKX5Myajcer45ySj6fFav4S56_Py_lDHlcgxRALaSFlNcM6pdKgUbyGsmQVBSVUZlNAyaoSaiFZWRtqTApZndkyKS3SUlVsSm4Pf3vX_ezQD3rd7VwbKjUDBopmWSiYEnpwVa7z3qHVvWu2xv1qoHrPTwd-es9PH_mFyM0h0iDiP7sUGaeK_QGfr2x8</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>3131908816</pqid></control><display><type>article</type><title>EvRepSL: Event-Stream Representation via Self-Supervised Learning for Event-Based Vision</title><source>IEEE Electronic Library (IEL)</source><creator>Qu, Qiang ; Chen, Xiaoming ; Ying Chung, Yuk ; Shen, Yiran</creator><creatorcontrib>Qu, Qiang ; Chen, Xiaoming ; Ying Chung, Yuk ; Shen, Yiran</creatorcontrib><description>Event-stream representation is the first step for many computer vision tasks using event cameras. It converts the asynchronous event-streams into a formatted structure so that conventional machine learning models can be applied easily. However, most of the state-of-the-art event-stream representations are manually designed and the quality of these representations cannot be guaranteed due to the noisy nature of event-streams. In this paper, we introduce a data-driven approach aiming at enhancing the quality of event-stream representations. Our approach commences with the introduction of a new event-stream representation based on spatial-temporal statistics, denoted as EvRep. Subsequently, we theoretically derive the intrinsic relationship between asynchronous event-streams and synchronous video frames. Building upon this theoretical relationship, we train a representation generator, RepGen, in a self-supervised learning manner accepting EvRep as input. Finally, the event-streams are converted to high-quality representations, termed as EvRepSL, by going through the learned RepGen (without the need of fine-tuning or retraining). Our methodology is rigorously validated through extensive evaluations on a variety of mainstream event-based classification and optical flow datasets (captured with various types of event cameras). The experimental results highlight not only our approach's superior performance over existing event-stream representations but also its versatility, being agnostic to different event cameras and tasks.</description><identifier>ISSN: 1057-7149</identifier><identifier>EISSN: 1941-0042</identifier><identifier>DOI: 10.1109/TIP.2024.3497795</identifier><identifier>CODEN: IIPRE4</identifier><language>eng</language><publisher>New York: IEEE</publisher><subject>Accuracy ; Cameras ; Computer vision ; Dynamic vision sensor ; Estimation ; event camera ; Event detection ; event-based vision ; Generators ; Machine learning ; neuromorphic vision ; Noise ; Noise reduction ; Optical flow ; Optical flow (image analysis) ; representation learning ; Representations ; Self-supervised learning ; Streams</subject><ispartof>IEEE transactions on image processing, 2024, Vol.33, p.6579-6591</ispartof><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c175t-57f163d3ed607aea94d1bb3c019598f61e73cb1d573bda0aa618d8fb2bfe0b9c3</cites><orcidid>0000-0002-6648-5050 ; 0000-0002-7503-3021 ; 0000-0002-3158-9650 ; 0000-0003-1385-1480</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/10758409$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>314,776,780,792,4010,27900,27901,27902,54733</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/10758409$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Qu, Qiang</creatorcontrib><creatorcontrib>Chen, Xiaoming</creatorcontrib><creatorcontrib>Ying Chung, Yuk</creatorcontrib><creatorcontrib>Shen, Yiran</creatorcontrib><title>EvRepSL: Event-Stream Representation via Self-Supervised Learning for Event-Based Vision</title><title>IEEE transactions on image processing</title><addtitle>TIP</addtitle><description>Event-stream representation is the first step for many computer vision tasks using event cameras. It converts the asynchronous event-streams into a formatted structure so that conventional machine learning models can be applied easily. However, most of the state-of-the-art event-stream representations are manually designed and the quality of these representations cannot be guaranteed due to the noisy nature of event-streams. In this paper, we introduce a data-driven approach aiming at enhancing the quality of event-stream representations. Our approach commences with the introduction of a new event-stream representation based on spatial-temporal statistics, denoted as EvRep. Subsequently, we theoretically derive the intrinsic relationship between asynchronous event-streams and synchronous video frames. Building upon this theoretical relationship, we train a representation generator, RepGen, in a self-supervised learning manner accepting EvRep as input. Finally, the event-streams are converted to high-quality representations, termed as EvRepSL, by going through the learned RepGen (without the need of fine-tuning or retraining). Our methodology is rigorously validated through extensive evaluations on a variety of mainstream event-based classification and optical flow datasets (captured with various types of event cameras). The experimental results highlight not only our approach's superior performance over existing event-stream representations but also its versatility, being agnostic to different event cameras and tasks.</description><subject>Accuracy</subject><subject>Cameras</subject><subject>Computer vision</subject><subject>Dynamic vision sensor</subject><subject>Estimation</subject><subject>event camera</subject><subject>Event detection</subject><subject>event-based vision</subject><subject>Generators</subject><subject>Machine learning</subject><subject>neuromorphic vision</subject><subject>Noise</subject><subject>Noise reduction</subject><subject>Optical flow</subject><subject>Optical flow (image analysis)</subject><subject>representation learning</subject><subject>Representations</subject><subject>Self-supervised learning</subject><subject>Streams</subject><issn>1057-7149</issn><issn>1941-0042</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2024</creationdate><recordtype>article</recordtype><sourceid>RIE</sourceid><recordid>eNpNkMFLwzAUxoMoOKd3Dx4KnjvzmqRpvOmYOigodoq3kLYv0rG1NekK_vdmbAdP7_G973sf_Ai5BjoDoOputXybJTThM8aVlEqckAkoDjGlPDkNOxUylsDVObnwfk0pcAHphHwtxnfsi_w-WozYDnExODTbKGgOfRDM0HRtNDYmKnBj42LXoxsbj3WUo3Ft035HtnPH8KPZHz4bHzKX5Myajcer45ySj6fFav4S56_Py_lDHlcgxRALaSFlNcM6pdKgUbyGsmQVBSVUZlNAyaoSaiFZWRtqTApZndkyKS3SUlVsSm4Pf3vX_ezQD3rd7VwbKjUDBopmWSiYEnpwVa7z3qHVvWu2xv1qoHrPTwd-es9PH_mFyM0h0iDiP7sUGaeK_QGfr2x8</recordid><startdate>2024</startdate><enddate>2024</enddate><creator>Qu, Qiang</creator><creator>Chen, Xiaoming</creator><creator>Ying Chung, Yuk</creator><creator>Shen, Yiran</creator><general>IEEE</general><general>The Institute of Electrical and Electronics Engineers, Inc. (IEEE)</general><scope>97E</scope><scope>RIA</scope><scope>RIE</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>7SP</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope><orcidid>https://orcid.org/0000-0002-6648-5050</orcidid><orcidid>https://orcid.org/0000-0002-7503-3021</orcidid><orcidid>https://orcid.org/0000-0002-3158-9650</orcidid><orcidid>https://orcid.org/0000-0003-1385-1480</orcidid></search><sort><creationdate>2024</creationdate><title>EvRepSL: Event-Stream Representation via Self-Supervised Learning for Event-Based Vision</title><author>Qu, Qiang ; Chen, Xiaoming ; Ying Chung, Yuk ; Shen, Yiran</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c175t-57f163d3ed607aea94d1bb3c019598f61e73cb1d573bda0aa618d8fb2bfe0b9c3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2024</creationdate><topic>Accuracy</topic><topic>Cameras</topic><topic>Computer vision</topic><topic>Dynamic vision sensor</topic><topic>Estimation</topic><topic>event camera</topic><topic>Event detection</topic><topic>event-based vision</topic><topic>Generators</topic><topic>Machine learning</topic><topic>neuromorphic vision</topic><topic>Noise</topic><topic>Noise reduction</topic><topic>Optical flow</topic><topic>Optical flow (image analysis)</topic><topic>representation learning</topic><topic>Representations</topic><topic>Self-supervised learning</topic><topic>Streams</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Qu, Qiang</creatorcontrib><creatorcontrib>Chen, Xiaoming</creatorcontrib><creatorcontrib>Ying Chung, Yuk</creatorcontrib><creatorcontrib>Shen, Yiran</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998-Present</collection><collection>IEEE Electronic Library (IEL)</collection><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Electronics & Communications Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>IEEE transactions on image processing</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Qu, Qiang</au><au>Chen, Xiaoming</au><au>Ying Chung, Yuk</au><au>Shen, Yiran</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>EvRepSL: Event-Stream Representation via Self-Supervised Learning for Event-Based Vision</atitle><jtitle>IEEE transactions on image processing</jtitle><stitle>TIP</stitle><date>2024</date><risdate>2024</risdate><volume>33</volume><spage>6579</spage><epage>6591</epage><pages>6579-6591</pages><issn>1057-7149</issn><eissn>1941-0042</eissn><coden>IIPRE4</coden><abstract>Event-stream representation is the first step for many computer vision tasks using event cameras. It converts the asynchronous event-streams into a formatted structure so that conventional machine learning models can be applied easily. However, most of the state-of-the-art event-stream representations are manually designed and the quality of these representations cannot be guaranteed due to the noisy nature of event-streams. In this paper, we introduce a data-driven approach aiming at enhancing the quality of event-stream representations. Our approach commences with the introduction of a new event-stream representation based on spatial-temporal statistics, denoted as EvRep. Subsequently, we theoretically derive the intrinsic relationship between asynchronous event-streams and synchronous video frames. Building upon this theoretical relationship, we train a representation generator, RepGen, in a self-supervised learning manner accepting EvRep as input. Finally, the event-streams are converted to high-quality representations, termed as EvRepSL, by going through the learned RepGen (without the need of fine-tuning or retraining). Our methodology is rigorously validated through extensive evaluations on a variety of mainstream event-based classification and optical flow datasets (captured with various types of event cameras). The experimental results highlight not only our approach's superior performance over existing event-stream representations but also its versatility, being agnostic to different event cameras and tasks.</abstract><cop>New York</cop><pub>IEEE</pub><doi>10.1109/TIP.2024.3497795</doi><tpages>13</tpages><orcidid>https://orcid.org/0000-0002-6648-5050</orcidid><orcidid>https://orcid.org/0000-0002-7503-3021</orcidid><orcidid>https://orcid.org/0000-0002-3158-9650</orcidid><orcidid>https://orcid.org/0000-0003-1385-1480</orcidid></addata></record> |
fulltext | fulltext_linktorsrc |
identifier | ISSN: 1057-7149 |
ispartof | IEEE transactions on image processing, 2024, Vol.33, p.6579-6591 |
issn | 1057-7149 1941-0042 |
language | eng |
recordid | cdi_proquest_journals_3131908816 |
source | IEEE Electronic Library (IEL) |
subjects | Accuracy Cameras Computer vision Dynamic vision sensor Estimation event camera Event detection event-based vision Generators Machine learning neuromorphic vision Noise Noise reduction Optical flow Optical flow (image analysis) representation learning Representations Self-supervised learning Streams |
title | EvRepSL: Event-Stream Representation via Self-Supervised Learning for Event-Based Vision |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-09T21%3A37%3A21IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_RIE&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=EvRepSL:%20Event-Stream%20Representation%20via%20Self-Supervised%20Learning%20for%20Event-Based%20Vision&rft.jtitle=IEEE%20transactions%20on%20image%20processing&rft.au=Qu,%20Qiang&rft.date=2024&rft.volume=33&rft.spage=6579&rft.epage=6591&rft.pages=6579-6591&rft.issn=1057-7149&rft.eissn=1941-0042&rft.coden=IIPRE4&rft_id=info:doi/10.1109/TIP.2024.3497795&rft_dat=%3Cproquest_RIE%3E3131908816%3C/proquest_RIE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=3131908816&rft_id=info:pmid/&rft_ieee_id=10758409&rfr_iscdi=true |