DeepForest: A Python package for RGB deep learning tree crown delineation

Remote sensing of forested landscapes can transform the speed, scale and cost of forest research. The delineation of individual trees in remote sensing images is an essential task in forest analysis. Here we introduce a new Python package, DeepForest that detects individual trees in high resolution...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Methods in ecology and evolution 2020-12, Vol.11 (12), p.1743-1751
Hauptverfasser: Weinstein, Ben G., Marconi, Sergio, Aubry‐Kientz, Mélaine, Vincent, Gregoire, Senyondo, Henry, White, Ethan P., Record, Sydne
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 1751
container_issue 12
container_start_page 1743
container_title Methods in ecology and evolution
container_volume 11
creator Weinstein, Ben G.
Marconi, Sergio
Aubry‐Kientz, Mélaine
Vincent, Gregoire
Senyondo, Henry
White, Ethan P.
Record, Sydne
description Remote sensing of forested landscapes can transform the speed, scale and cost of forest research. The delineation of individual trees in remote sensing images is an essential task in forest analysis. Here we introduce a new Python package, DeepForest that detects individual trees in high resolution RGB imagery using deep learning. While deep learning has proven highly effective in a range of computer vision tasks, it requires large amounts of training data that are typically difficult to obtain in ecological studies. DeepForest overcomes this limitation by including a model pretrained on over 30 million algorithmically generated crowns from 22 forests and fine‐tuned using 10,000 hand‐labelled crowns from six forests. The package supports the application of this general model to new data, fine tuning the model to new datasets with user labelled crowns, training new models and evaluating model predictions. This simplifies the process of using and retraining deep learning models for a range of forests, sensors and spatial resolutions. We illustrate the workflow of DeepForest using data from the National Ecological Observatory Network, a tropical forest in French Guiana, and street trees from Portland, Oregon.
doi_str_mv 10.1111/2041-210X.13472
format Article
fullrecord <record><control><sourceid>proquest_hal_p</sourceid><recordid>TN_cdi_hal_primary_oai_HAL_hal_03073497v1</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2466150122</sourcerecordid><originalsourceid>FETCH-LOGICAL-c4572-976e1905581bd2417f653cf073b13183444d82a949f5f5b68bba33339f775edb3</originalsourceid><addsrcrecordid>eNqFkM1LwzAYxoMoOObOXgOePHTLZ9N6m3NfMFFEwVtI22TrrE1Nu43996ZWhjefS8Kb3_OQ9wHgGqMh9hoRxHBAMHofYsoEOQO90-T8z_0SDOp6i7xoFCPCemD5oHU1s07XzR0cw-djs7ElrFT6odYaGuvgy_weZh6ChVauzMs1bJzWMHX2UPqHIi-1anJbXoELo4paD37PPnibTV8ni2D1NF9OxqsgZVyQIBahxjHiPMJJRhgWJuQ0NUjQBFMcUcZYFhEVs9hww5MwShJFvWIjBNdZQvvgtsvdqEJWLv9U7iityuVivJLtDFEfxmKxx5696djK2a-d31Fu7c6V_nuSsDDEHGFCPDXqKL9TXTttTrEYybZe2RYo2wLlT73eEXaOQ17o43-4fJxOaWf8BkRseKw</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2466150122</pqid></control><display><type>article</type><title>DeepForest: A Python package for RGB deep learning tree crown delineation</title><source>Access via Wiley Online Library</source><source>Alma/SFX Local Collection</source><creator>Weinstein, Ben G. ; Marconi, Sergio ; Aubry‐Kientz, Mélaine ; Vincent, Gregoire ; Senyondo, Henry ; White, Ethan P. ; Record, Sydne</creator><contributor>Record, Sydne</contributor><creatorcontrib>Weinstein, Ben G. ; Marconi, Sergio ; Aubry‐Kientz, Mélaine ; Vincent, Gregoire ; Senyondo, Henry ; White, Ethan P. ; Record, Sydne ; Record, Sydne</creatorcontrib><description>Remote sensing of forested landscapes can transform the speed, scale and cost of forest research. The delineation of individual trees in remote sensing images is an essential task in forest analysis. Here we introduce a new Python package, DeepForest that detects individual trees in high resolution RGB imagery using deep learning. While deep learning has proven highly effective in a range of computer vision tasks, it requires large amounts of training data that are typically difficult to obtain in ecological studies. DeepForest overcomes this limitation by including a model pretrained on over 30 million algorithmically generated crowns from 22 forests and fine‐tuned using 10,000 hand‐labelled crowns from six forests. The package supports the application of this general model to new data, fine tuning the model to new datasets with user labelled crowns, training new models and evaluating model predictions. This simplifies the process of using and retraining deep learning models for a range of forests, sensors and spatial resolutions. We illustrate the workflow of DeepForest using data from the National Ecological Observatory Network, a tropical forest in French Guiana, and street trees from Portland, Oregon.</description><identifier>ISSN: 2041-210X</identifier><identifier>EISSN: 2041-210X</identifier><identifier>DOI: 10.1111/2041-210X.13472</identifier><language>eng</language><publisher>London: John Wiley &amp; Sons, Inc</publisher><subject>Biodiversity ; Biodiversity and Ecology ; Botanics ; Computer vision ; crown delineation ; Deep learning ; Delineation ; Ecological studies ; Ecology, environment ; Ecosystems ; Environmental Sciences ; Forestry research ; Forests ; Image resolution ; Life Sciences ; NEON ; Remote sensing ; Retraining ; RGB ; Systematics, Phylogenetics and taxonomy ; Training ; tree crowns ; Trees ; Tropical forests ; Vegetal Biology ; Workflow</subject><ispartof>Methods in ecology and evolution, 2020-12, Vol.11 (12), p.1743-1751</ispartof><rights>2020 British Ecological Society</rights><rights>Methods in Ecology and Evolution © 2020 British Ecological Society</rights><rights>Attribution</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c4572-976e1905581bd2417f653cf073b13183444d82a949f5f5b68bba33339f775edb3</citedby><cites>FETCH-LOGICAL-c4572-976e1905581bd2417f653cf073b13183444d82a949f5f5b68bba33339f775edb3</cites><orcidid>0000-0001-6728-7745 ; 0000-0002-2176-7935</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktopdf>$$Uhttps://onlinelibrary.wiley.com/doi/pdf/10.1111%2F2041-210X.13472$$EPDF$$P50$$Gwiley$$H</linktopdf><linktohtml>$$Uhttps://onlinelibrary.wiley.com/doi/full/10.1111%2F2041-210X.13472$$EHTML$$P50$$Gwiley$$H</linktohtml><link.rule.ids>230,314,780,784,885,1417,27924,27925,45574,45575</link.rule.ids><backlink>$$Uhttps://hal.inrae.fr/hal-03073497$$DView record in HAL$$Hfree_for_read</backlink></links><search><contributor>Record, Sydne</contributor><creatorcontrib>Weinstein, Ben G.</creatorcontrib><creatorcontrib>Marconi, Sergio</creatorcontrib><creatorcontrib>Aubry‐Kientz, Mélaine</creatorcontrib><creatorcontrib>Vincent, Gregoire</creatorcontrib><creatorcontrib>Senyondo, Henry</creatorcontrib><creatorcontrib>White, Ethan P.</creatorcontrib><creatorcontrib>Record, Sydne</creatorcontrib><title>DeepForest: A Python package for RGB deep learning tree crown delineation</title><title>Methods in ecology and evolution</title><description>Remote sensing of forested landscapes can transform the speed, scale and cost of forest research. The delineation of individual trees in remote sensing images is an essential task in forest analysis. Here we introduce a new Python package, DeepForest that detects individual trees in high resolution RGB imagery using deep learning. While deep learning has proven highly effective in a range of computer vision tasks, it requires large amounts of training data that are typically difficult to obtain in ecological studies. DeepForest overcomes this limitation by including a model pretrained on over 30 million algorithmically generated crowns from 22 forests and fine‐tuned using 10,000 hand‐labelled crowns from six forests. The package supports the application of this general model to new data, fine tuning the model to new datasets with user labelled crowns, training new models and evaluating model predictions. This simplifies the process of using and retraining deep learning models for a range of forests, sensors and spatial resolutions. We illustrate the workflow of DeepForest using data from the National Ecological Observatory Network, a tropical forest in French Guiana, and street trees from Portland, Oregon.</description><subject>Biodiversity</subject><subject>Biodiversity and Ecology</subject><subject>Botanics</subject><subject>Computer vision</subject><subject>crown delineation</subject><subject>Deep learning</subject><subject>Delineation</subject><subject>Ecological studies</subject><subject>Ecology, environment</subject><subject>Ecosystems</subject><subject>Environmental Sciences</subject><subject>Forestry research</subject><subject>Forests</subject><subject>Image resolution</subject><subject>Life Sciences</subject><subject>NEON</subject><subject>Remote sensing</subject><subject>Retraining</subject><subject>RGB</subject><subject>Systematics, Phylogenetics and taxonomy</subject><subject>Training</subject><subject>tree crowns</subject><subject>Trees</subject><subject>Tropical forests</subject><subject>Vegetal Biology</subject><subject>Workflow</subject><issn>2041-210X</issn><issn>2041-210X</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><recordid>eNqFkM1LwzAYxoMoOObOXgOePHTLZ9N6m3NfMFFEwVtI22TrrE1Nu43996ZWhjefS8Kb3_OQ9wHgGqMh9hoRxHBAMHofYsoEOQO90-T8z_0SDOp6i7xoFCPCemD5oHU1s07XzR0cw-djs7ElrFT6odYaGuvgy_weZh6ChVauzMs1bJzWMHX2UPqHIi-1anJbXoELo4paD37PPnibTV8ni2D1NF9OxqsgZVyQIBahxjHiPMJJRhgWJuQ0NUjQBFMcUcZYFhEVs9hww5MwShJFvWIjBNdZQvvgtsvdqEJWLv9U7iityuVivJLtDFEfxmKxx5696djK2a-d31Fu7c6V_nuSsDDEHGFCPDXqKL9TXTttTrEYybZe2RYo2wLlT73eEXaOQ17o43-4fJxOaWf8BkRseKw</recordid><startdate>202012</startdate><enddate>202012</enddate><creator>Weinstein, Ben G.</creator><creator>Marconi, Sergio</creator><creator>Aubry‐Kientz, Mélaine</creator><creator>Vincent, Gregoire</creator><creator>Senyondo, Henry</creator><creator>White, Ethan P.</creator><creator>Record, Sydne</creator><general>John Wiley &amp; Sons, Inc</general><general>Wiley</general><scope>AAYXX</scope><scope>CITATION</scope><scope>7QG</scope><scope>7SN</scope><scope>8FD</scope><scope>C1K</scope><scope>FR3</scope><scope>P64</scope><scope>RC3</scope><scope>1XC</scope><scope>VOOES</scope><orcidid>https://orcid.org/0000-0001-6728-7745</orcidid><orcidid>https://orcid.org/0000-0002-2176-7935</orcidid></search><sort><creationdate>202012</creationdate><title>DeepForest: A Python package for RGB deep learning tree crown delineation</title><author>Weinstein, Ben G. ; Marconi, Sergio ; Aubry‐Kientz, Mélaine ; Vincent, Gregoire ; Senyondo, Henry ; White, Ethan P. ; Record, Sydne</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c4572-976e1905581bd2417f653cf073b13183444d82a949f5f5b68bba33339f775edb3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Biodiversity</topic><topic>Biodiversity and Ecology</topic><topic>Botanics</topic><topic>Computer vision</topic><topic>crown delineation</topic><topic>Deep learning</topic><topic>Delineation</topic><topic>Ecological studies</topic><topic>Ecology, environment</topic><topic>Ecosystems</topic><topic>Environmental Sciences</topic><topic>Forestry research</topic><topic>Forests</topic><topic>Image resolution</topic><topic>Life Sciences</topic><topic>NEON</topic><topic>Remote sensing</topic><topic>Retraining</topic><topic>RGB</topic><topic>Systematics, Phylogenetics and taxonomy</topic><topic>Training</topic><topic>tree crowns</topic><topic>Trees</topic><topic>Tropical forests</topic><topic>Vegetal Biology</topic><topic>Workflow</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Weinstein, Ben G.</creatorcontrib><creatorcontrib>Marconi, Sergio</creatorcontrib><creatorcontrib>Aubry‐Kientz, Mélaine</creatorcontrib><creatorcontrib>Vincent, Gregoire</creatorcontrib><creatorcontrib>Senyondo, Henry</creatorcontrib><creatorcontrib>White, Ethan P.</creatorcontrib><creatorcontrib>Record, Sydne</creatorcontrib><collection>CrossRef</collection><collection>Animal Behavior Abstracts</collection><collection>Ecology Abstracts</collection><collection>Technology Research Database</collection><collection>Environmental Sciences and Pollution Management</collection><collection>Engineering Research Database</collection><collection>Biotechnology and BioEngineering Abstracts</collection><collection>Genetics Abstracts</collection><collection>Hyper Article en Ligne (HAL)</collection><collection>Hyper Article en Ligne (HAL) (Open Access)</collection><jtitle>Methods in ecology and evolution</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Weinstein, Ben G.</au><au>Marconi, Sergio</au><au>Aubry‐Kientz, Mélaine</au><au>Vincent, Gregoire</au><au>Senyondo, Henry</au><au>White, Ethan P.</au><au>Record, Sydne</au><au>Record, Sydne</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>DeepForest: A Python package for RGB deep learning tree crown delineation</atitle><jtitle>Methods in ecology and evolution</jtitle><date>2020-12</date><risdate>2020</risdate><volume>11</volume><issue>12</issue><spage>1743</spage><epage>1751</epage><pages>1743-1751</pages><issn>2041-210X</issn><eissn>2041-210X</eissn><abstract>Remote sensing of forested landscapes can transform the speed, scale and cost of forest research. The delineation of individual trees in remote sensing images is an essential task in forest analysis. Here we introduce a new Python package, DeepForest that detects individual trees in high resolution RGB imagery using deep learning. While deep learning has proven highly effective in a range of computer vision tasks, it requires large amounts of training data that are typically difficult to obtain in ecological studies. DeepForest overcomes this limitation by including a model pretrained on over 30 million algorithmically generated crowns from 22 forests and fine‐tuned using 10,000 hand‐labelled crowns from six forests. The package supports the application of this general model to new data, fine tuning the model to new datasets with user labelled crowns, training new models and evaluating model predictions. This simplifies the process of using and retraining deep learning models for a range of forests, sensors and spatial resolutions. We illustrate the workflow of DeepForest using data from the National Ecological Observatory Network, a tropical forest in French Guiana, and street trees from Portland, Oregon.</abstract><cop>London</cop><pub>John Wiley &amp; Sons, Inc</pub><doi>10.1111/2041-210X.13472</doi><tpages>9</tpages><orcidid>https://orcid.org/0000-0001-6728-7745</orcidid><orcidid>https://orcid.org/0000-0002-2176-7935</orcidid><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier ISSN: 2041-210X
ispartof Methods in ecology and evolution, 2020-12, Vol.11 (12), p.1743-1751
issn 2041-210X
2041-210X
language eng
recordid cdi_hal_primary_oai_HAL_hal_03073497v1
source Access via Wiley Online Library; Alma/SFX Local Collection
subjects Biodiversity
Biodiversity and Ecology
Botanics
Computer vision
crown delineation
Deep learning
Delineation
Ecological studies
Ecology, environment
Ecosystems
Environmental Sciences
Forestry research
Forests
Image resolution
Life Sciences
NEON
Remote sensing
Retraining
RGB
Systematics, Phylogenetics and taxonomy
Training
tree crowns
Trees
Tropical forests
Vegetal Biology
Workflow
title DeepForest: A Python package for RGB deep learning tree crown delineation
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T14%3A08%3A28IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_hal_p&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=DeepForest:%20A%20Python%20package%20for%20RGB%20deep%20learning%20tree%20crown%20delineation&rft.jtitle=Methods%20in%20ecology%20and%20evolution&rft.au=Weinstein,%20Ben%20G.&rft.date=2020-12&rft.volume=11&rft.issue=12&rft.spage=1743&rft.epage=1751&rft.pages=1743-1751&rft.issn=2041-210X&rft.eissn=2041-210X&rft_id=info:doi/10.1111/2041-210X.13472&rft_dat=%3Cproquest_hal_p%3E2466150122%3C/proquest_hal_p%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2466150122&rft_id=info:pmid/&rfr_iscdi=true