Extending scene-to-patch models: Multi-resolution multiple instance learning for Earth observation
Abstract Land cover classification (LCC) and natural disaster response (NDR) are important issues in climate change mitigation and adaptation. Existing approaches that use machine learning with Earth observation (EO) imaging data for LCC and NDR often rely on fully annotated and segmented datasets....
Saved in:
Published in | Environmental Data Science Vol. 2 |
---|---|
Main Authors | , , , |
Format | Journal Article |
Language | English |
Published |
Cambridge University Press
2023
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Abstract | Abstract
Land cover classification (LCC) and natural disaster response (NDR) are important issues in climate change mitigation and adaptation. Existing approaches that use machine learning with Earth observation (EO) imaging data for LCC and NDR often rely on fully annotated and segmented datasets. Creating these datasets requires a large amount of effort, and a lack of suitable datasets has become an obstacle in scaling the use of machine learning for EO. In this study, we extend our prior work on Scene-to-Patch models: an alternative machine learning approach for EO that utilizes Multiple Instance Learning (MIL). As our approach only requires high-level scene labels, it enables much faster development of new datasets while still providing segmentation through patch-level predictions, ultimately increasing the accessibility of using machine learning for EO. We propose new multi-resolution MIL architectures that outperform single-resolution MIL models and non-MIL baselines on the DeepGlobe LCC and FloodNet NDR datasets. In addition, we conduct a thorough analysis of model performance and interpretability. |
---|---|
AbstractList | Land cover classification (LCC) and natural disaster response (NDR) are important issues in climate change mitigation and adaptation. Existing approaches that use machine learning with Earth observation (EO) imaging data for LCC and NDR often rely on fully annotated and segmented datasets. Creating these datasets requires a large amount of effort, and a lack of suitable datasets has become an obstacle in scaling the use of machine learning for EO. In this study, we extend our prior work on Scene-to-Patch models: an alternative machine learning approach for EO that utilizes Multiple Instance Learning (MIL). As our approach only requires high-level scene labels, it enables much faster development of new datasets while still providing segmentation through patch-level predictions, ultimately increasing the accessibility of using machine learning for EO. We propose new multi-resolution MIL architectures that outperform single-resolution MIL models and non-MIL baselines on the DeepGlobe LCC and FloodNet NDR datasets. In addition, we conduct a thorough analysis of model performance and interpretability. Abstract Land cover classification (LCC) and natural disaster response (NDR) are important issues in climate change mitigation and adaptation. Existing approaches that use machine learning with Earth observation (EO) imaging data for LCC and NDR often rely on fully annotated and segmented datasets. Creating these datasets requires a large amount of effort, and a lack of suitable datasets has become an obstacle in scaling the use of machine learning for EO. In this study, we extend our prior work on Scene-to-Patch models: an alternative machine learning approach for EO that utilizes Multiple Instance Learning (MIL). As our approach only requires high-level scene labels, it enables much faster development of new datasets while still providing segmentation through patch-level predictions, ultimately increasing the accessibility of using machine learning for EO. We propose new multi-resolution MIL architectures that outperform single-resolution MIL models and non-MIL baselines on the DeepGlobe LCC and FloodNet NDR datasets. In addition, we conduct a thorough analysis of model performance and interpretability. |
ArticleNumber | e42 |
Author | Early, Joseph Evers, Christine Ramchurn, Sarvapali Deweese, Ying-Jung Chen |
Author_xml | – sequence: 1 givenname: Joseph orcidid: 0000-0001-7748-9340 surname: Early fullname: Early, Joseph – sequence: 2 givenname: Ying-Jung Chen orcidid: 0000-0002-5946-7831 surname: Deweese fullname: Deweese, Ying-Jung Chen – sequence: 3 givenname: Christine surname: Evers fullname: Evers, Christine – sequence: 4 givenname: Sarvapali surname: Ramchurn fullname: Ramchurn, Sarvapali |
BookMark | eNpN0E1LAzEQBuAgCtbak38gd9maz03Wm5SqhYqX3kM2mW23bJOSbEX_vbtWxNMML8zD8N6gyxADIHRHyZwSqh7A5zkjjM85uUATVnJRiJKwy3_7NZrlvCeEcEaZEnKC6uVnD8G3YYuzgwBFH4uj7d0OH6KHLj_it1PXt0WCHLtT38aAD2Nw7AC3Ifc2OMAd2BRGookJL23qdzjWGdKHHQ9u0VVjuwyz3zlFm-flZvFarN9fVoundeEYE33hSye90rW2VvqaqlqB5KXimhEvNXVCetDgFBEarAfZ1ErwRjUSRFlXlk_R6sz6aPfmmNqDTV8m2tb8BDFtzfBZ6zowqgIJFTDa6MHgXhPNQVUVcZRyxWGw7s-WSzHnBM2fR4kZyzZD2WYs23DCvwGGk3Vc |
Cites_doi | 10.3389/fenvs.2019.00127 10.1016/j.patcog.2017.08.026 10.3390/rs12183053 10.3390/rs12101667 10.1109/TGRS.2011.2171691 10.3390/w11122499 10.1038/s41558-022-01377-7 10.1007/978-3-319-24574-4_28 10.1073/pnas.2018863118 10.1007/s11263-009-0275-4 10.1145/3485128 10.3390/rs12020207 10.1109/TGRS.2017.2750220 10.1016/j.rse.2019.111322 10.3390/rs14205095 10.3390/en15155522 10.3390/rs11030237 10.1109/MWENT47943.2020.9067475 10.1016/j.ijinfomgt.2019.102049 10.5194/essd-12-3269-2020 10.1109/ACCESS.2021.3090981 10.3390/rs11050531 10.1016/j.ecolind.2022.109341 10.3390/su13147547 10.1016/j.artmed.2019.101744 10.1038/s41598-019-56967-x 10.1016/j.aei.2019.101009 10.1016/j.patcog.2017.10.009 10.1109/TPAMI.2021.3059968 10.1109/LGRS.2020.3007183 10.1016/j.ijinfomgt.2019.04.004 10.1609/aaai.v33i01.3301702 10.1007/978-3-030-69544-6_27 10.1111/gcb.15962 |
ContentType | Journal Article |
DBID | AAYXX CITATION DOA |
DOI | 10.1017/eds.2023.30 |
DatabaseName | CrossRef Directory of Open Access Journals |
DatabaseTitle | CrossRef |
DatabaseTitleList | CrossRef |
Database_xml | – sequence: 1 dbid: DOA name: Directory of Open Access Journals url: https://www.doaj.org/ sourceTypes: Open Website |
DeliveryMethod | fulltext_linktorsrc |
EISSN | 2634-4602 |
ExternalDocumentID | oai_doaj_org_article_79e5e9e21f87433d8083e7990c11373e 10_1017_eds_2023_30 |
GroupedDBID | 09C 09E 0R~ AANRG AASVR AAYXX ABVZP ACFCP ACZWT ADDNB ADKIL ADVJH AEBAK AFRIC AGABE AGBYD AGJUD AHIPN AHRGI ALMA_UNASSIGNED_HOLDINGS AQJOH BLZWO CCQAD CITATION CJCSC GROUPED_DOAJ IKXGN IPYYG OK1 RCA ROL WFFJZ |
ID | FETCH-LOGICAL-c224t-d6c5d78b8aa5db17b7e53673820d581c45de8ec7048eade5fb743f7f5e46b9a3 |
IEDL.DBID | DOA |
ISSN | 2634-4602 |
IngestDate | Mon Sep 09 19:22:14 EDT 2024 Fri Aug 23 00:54:21 EDT 2024 |
IsDoiOpenAccess | true |
IsOpenAccess | true |
IsPeerReviewed | true |
IsScholarly | true |
Language | English |
LinkModel | DirectLink |
MergedId | FETCHMERGED-LOGICAL-c224t-d6c5d78b8aa5db17b7e53673820d581c45de8ec7048eade5fb743f7f5e46b9a3 |
ORCID | 0000-0001-7748-9340 0000-0002-5946-7831 |
OpenAccessLink | https://doaj.org/article/79e5e9e21f87433d8083e7990c11373e |
ParticipantIDs | doaj_primary_oai_doaj_org_article_79e5e9e21f87433d8083e7990c11373e crossref_primary_10_1017_eds_2023_30 |
PublicationCentury | 2000 |
PublicationDate | 2023-00-00 2023-01-01 |
PublicationDateYYYYMMDD | 2023-01-01 |
PublicationDate_xml | – year: 2023 text: 2023-00-00 |
PublicationDecade | 2020 |
PublicationTitle | Environmental Data Science |
PublicationYear | 2023 |
Publisher | Cambridge University Press |
Publisher_xml | – name: Cambridge University Press |
References | Hashimoto (S2634460223000304_r17) 2020 S2634460223000304_r7 S2634460223000304_r20 S2634460223000304_r8 S2634460223000304_r21 S2634460223000304_r22 S2634460223000304_r24 S2634460223000304_r26 S2634460223000304_r28 Rakhlin (S2634460223000304_r38) 2018 S2634460223000304_r50 S2634460223000304_r51 S2634460223000304_r52 S2634460223000304_r53 S2634460223000304_r54 S2634460223000304_r55 S2634460223000304_r11 S2634460223000304_r56 S2634460223000304_r12 S2634460223000304_r13 S2634460223000304_r57 S2634460223000304_r14 S2634460223000304_r15 S2634460223000304_r16 S2634460223000304_r19 S2634460223000304_r5 Seferbekov (S2634460223000304_r44) 2018 S2634460223000304_r6 Vaswani (S2634460223000304_r48) 2017; 30 Early (S2634460223000304_r10) 2022a; 35 S2634460223000304_r2 S2634460223000304_r40 S2634460223000304_r41 S2634460223000304_r42 S2634460223000304_r43 S2634460223000304_r45 S2634460223000304_r46 S2634460223000304_r47 Robinson (S2634460223000304_r39) 2019 Vatsavai (S2634460223000304_r49) 2013; 2013 Long (S2634460223000304_r31) 2015 S2634460223000304_r30 S2634460223000304_r32 S2634460223000304_r33 S2634460223000304_r34 S2634460223000304_r35 S2634460223000304_r36 S2634460223000304_r37 Lin (S2634460223000304_r29) 2017 Karra (S2634460223000304_r23) 2021 Kuo (S2634460223000304_r25) 2018 He (S2634460223000304_r18) 2016 Li (S2634460223000304_r27) 2021 Demir (S2634460223000304_r9) 2018 |
References_xml | – ident: S2634460223000304_r11 – volume-title: 2020 IEEE/CVF Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). year: 2020 ident: S2634460223000304_r17 contributor: fullname: Hashimoto – ident: S2634460223000304_r35 doi: 10.3389/fenvs.2019.00127 – ident: S2634460223000304_r53 doi: 10.1016/j.patcog.2017.08.026 – ident: S2634460223000304_r19 doi: 10.3390/rs12183053 – ident: S2634460223000304_r20 doi: 10.3390/rs12101667 – volume-title: 2018 IEEE/CVF Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) year: 2018 ident: S2634460223000304_r25 contributor: fullname: Kuo – ident: S2634460223000304_r55 doi: 10.1109/TGRS.2011.2171691 – ident: S2634460223000304_r6 doi: 10.3390/w11122499 – volume: 35 start-page: 27652 year: 2022a ident: S2634460223000304_r10 article-title: Non-Markovian reward modelling from trajectory labels via interpretable multiple instance learning publication-title: Advances in Neural Information Processing Systems contributor: fullname: Early – ident: S2634460223000304_r22 doi: 10.1038/s41558-022-01377-7 – ident: S2634460223000304_r41 doi: 10.1007/978-3-319-24574-4_28 – ident: S2634460223000304_r26 doi: 10.1073/pnas.2018863118 – ident: S2634460223000304_r12 – ident: S2634460223000304_r14 doi: 10.1007/s11263-009-0275-4 – ident: S2634460223000304_r40 doi: 10.1145/3485128 – ident: S2634460223000304_r51 doi: 10.3390/rs12020207 – volume: 30 start-page: 5998 year: 2017 ident: S2634460223000304_r48 article-title: Attention is all you need publication-title: Advances in Neural Information Processing Systems contributor: fullname: Vaswani – volume-title: 2018 IEEE/CVF Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) year: 2018 ident: S2634460223000304_r38 contributor: fullname: Rakhlin – volume-title: 2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS year: 2021 ident: S2634460223000304_r23 contributor: fullname: Karra – ident: S2634460223000304_r30 doi: 10.1109/TGRS.2017.2750220 – ident: S2634460223000304_r45 doi: 10.1016/j.rse.2019.111322 – ident: S2634460223000304_r52 doi: 10.3390/rs14205095 – volume-title: 2018 IEEE/CVF Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) year: 2018 ident: S2634460223000304_r9 contributor: fullname: Demir – ident: S2634460223000304_r43 doi: 10.3390/en15155522 – volume-title: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) year: 2015 ident: S2634460223000304_r31 contributor: fullname: Long – ident: S2634460223000304_r5 doi: 10.3390/rs11030237 – ident: S2634460223000304_r24 doi: 10.1109/MWENT47943.2020.9067475 – ident: S2634460223000304_r32 – ident: S2634460223000304_r15 doi: 10.1016/j.ijinfomgt.2019.102049 – volume-title: 2018 IEEE/CVF Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) year: 2018 ident: S2634460223000304_r44 contributor: fullname: Seferbekov – ident: S2634460223000304_r16 doi: 10.5194/essd-12-3269-2020 – ident: S2634460223000304_r37 doi: 10.1109/ACCESS.2021.3090981 – ident: S2634460223000304_r54 doi: 10.3390/rs11050531 – ident: S2634460223000304_r47 doi: 10.1016/j.ecolind.2022.109341 – ident: S2634460223000304_r34 doi: 10.3390/su13147547 – volume: 2013 year: 2013 ident: S2634460223000304_r49 article-title: Complex settlement pattern extraction with multi-instance learning publication-title: Joint Urban Remote Sensing Event contributor: fullname: Vatsavai – ident: S2634460223000304_r28 doi: 10.1016/j.artmed.2019.101744 – volume-title: 2017 IEEE Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) year: 2017 ident: S2634460223000304_r29 contributor: fullname: Lin – ident: S2634460223000304_r2 doi: 10.1038/s41598-019-56967-x – volume-title: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) year: 2019 ident: S2634460223000304_r39 contributor: fullname: Robinson – ident: S2634460223000304_r36 doi: 10.1016/j.aei.2019.101009 – ident: S2634460223000304_r7 doi: 10.1016/j.patcog.2017.10.009 – ident: S2634460223000304_r8 – volume-title: 2016 IEEE Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) year: 2016 ident: S2634460223000304_r18 contributor: fullname: He – volume-title: 2021 Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) year: 2021 ident: S2634460223000304_r27 contributor: fullname: Li – ident: S2634460223000304_r33 doi: 10.1109/TPAMI.2021.3059968 – ident: S2634460223000304_r21 – ident: S2634460223000304_r57 doi: 10.1109/LGRS.2020.3007183 – ident: S2634460223000304_r46 – ident: S2634460223000304_r56 doi: 10.1016/j.ijinfomgt.2019.04.004 – ident: S2634460223000304_r42 doi: 10.1609/aaai.v33i01.3301702 – ident: S2634460223000304_r50 doi: 10.1007/978-3-030-69544-6_27 – ident: S2634460223000304_r13 doi: 10.1111/gcb.15962 |
SSID | ssj0003212745 |
Score | 2.2612538 |
Snippet | Abstract
Land cover classification (LCC) and natural disaster response (NDR) are important issues in climate change mitigation and adaptation. Existing... Land cover classification (LCC) and natural disaster response (NDR) are important issues in climate change mitigation and adaptation. Existing approaches that... |
SourceID | doaj crossref |
SourceType | Open Website Aggregation Database |
SubjectTerms | Earth observation multi-resolution multiple instance learning remote sensing scene-to-patch models |
Title | Extending scene-to-patch models: Multi-resolution multiple instance learning for Earth observation |
URI | https://doaj.org/article/79e5e9e21f87433d8083e7990c11373e |
Volume | 2 |
hasFullText | 1 |
inHoldings | 1 |
isFullTextHit | |
isPrint | |
link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwrV09T8MwELVQJxYEAsS3PHQ11HEcO2yAUlUMTEXqZvnjTFnSCsL_5-ykVTYW1siKnHfOvbvT-R0hU4V5rAXQzBZWM2Q8yVylFKuqAMKXIQaXuy3eqsV7-bqSq9Gor9QT1ssD98A9qBok1FDwqJHsRNAYM4BCH-o5F0pA9r5cjpKp5INFEi4v5XAhL2lEQ0ji3IW4T_3OIwoaKfVnSpkfk6MhFqRP_R5OyAG0p8Q1uSyNjEKT0BKwbsO26C_XNA-t-X6k-c4swzR5ODV01xRIP3Os54EOsyA-KIaktMGPXNON29dfz8hy3ixfFmwYhMA8MmzHQuVlUNppa2VwXDkFUqRxncUsSM19KQNo8Ar_xtT_LKNDqKKKEsrK1Vack0m7aeGC0BAdQmdjwEwPqVy4oLmQM6Eipk0YOVyS6Q4as-3lLkzfB6YMImgSgkbgsucE235J0qjOD9ByZrCc-ctyV__xkmtymPbUF0VuyKT7-oFbDBM6d5dPxC946Lu0 |
link.rule.ids | 315,786,790,870,2115,4043,27954,27955,27956 |
linkProvider | Directory of Open Access Journals |
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%3Ajournal&rft.genre=article&rft.atitle=Extending+scene-to-patch+models%3A+Multi-resolution+multiple+instance+learning+for+Earth+observation&rft.jtitle=Environmental+Data+Science&rft.au=Early%2C+Joseph&rft.au=Deweese%2C+Ying-Jung+Chen&rft.au=Evers%2C+Christine&rft.au=Ramchurn%2C+Sarvapali&rft.date=2023&rft.issn=2634-4602&rft.eissn=2634-4602&rft.volume=2&rft_id=info:doi/10.1017%2Feds.2023.30&rft.externalDBID=n%2Fa&rft.externalDocID=10_1017_eds_2023_30 |
thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=2634-4602&client=summon |
thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=2634-4602&client=summon |
thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=2634-4602&client=summon |