{"refrec":{"BRefID":347969,"RR":"<b>Hoekendijk, J.P.A.; Kellenberger, B.; Aarts, G.M; Brasseur, S.; Poiesz, S.; Tuia, D.</b> (2021). Counting using deep learning regression gives value to ecological surveys. <i>NPG Scientific Reports 11</i>: 23209. <a href=\"https://dx.doi.org/10.1038/s41598-021-02387-9\" target=\"_blank\">https://dx.doi.org/10.1038/s41598-021-02387-9</a>","BEntID":344661,"PublicFlag":1,"CheckedFlag":0,"wosflag":1,"vabbflag":1,"RefStringPartII":". <i>NPG Scientific Reports 11</i>: 23209. <a href=\"https://dx.doi.org/10.1038/s41598-021-02387-9\" target=\"_blank\">https://dx.doi.org/10.1038/s41598-021-02387-9</a>","DocTypID":8,"DocType":"Journal article","MarineFlag":0,"FreshFlag":0,"BrackishFlag":0,"TerrestrialFlag":0,"Authorstring":"Hoekendijk, J.P.A.; Kellenberger, B.; Aarts, G.M; Brasseur, S.; Poiesz, S.; Tuia, D.","OrigTitleTranslFlag":0,"Authorstringtrunc":"Hoekendijk, J.P.A. <i>et al.</i>","Englishabstract":"<p>    Many ecological studies rely on count data and involve manual counting of    objects of interest, which is time-consuming and especially disadvantageous    when time in the field or lab is limited. However, an increasing number of    works uses digital imagery, which opens opportunities to automatise    counting tasks. In this study, we use machine learning to automate counting    objects of interest without the need to label individual objects. By    leveraging already existing image-level annotations, this approach can also    give value to historical data that were collected and annotated over longer    time series (typical for many ecological studies), without the aim of deep    learning applications. We demonstrate deep learning regression on two    fundamentally different counting tasks: (i) daily growth rings from    microscopic images of fish otolith (i.e., hearing stone) and (ii) hauled    out seals from highly variable aerial imagery. In the otolith images, our    deep learning-based regressor yields an <em>RMSE</em> of 3.40 day-rings andan R2R2 of 0.92. Initial performance in the seal images is lower (    <em>RMSE</em> of 23.46 seals and R2R2 of 0.72), which can be attributed to    a lack of images with a high number of seals in the initial training set,    compared to the test set. We then show how to improve performance    substantially (<em>RMSE</em> of 19.03 seals and R2R2 of 0.77) by carefully    selecting and relabelling just 100 additional training images based on    initial model prediction discrepancy. The regression-based approach used    here returns accurate counts (R2R2 of 0.92 and 0.77 for the rings and    seals, respectively), directly usable in ecological research.</p>","AbstractOtherLang":null,"BibLvlCode":"AS","StandardTitle":"Counting using deep learning regression gives value to ecological surveys","OrigTitleLangCode":"en","OrigTitleLangCodeExtended":"eng","OrigTitleLangID":15,"DateLastModified":{"date":"2026-04-20 01:32:48.297392","timezone_type":1,"timezone":"+02:00"},"UserAccessRight":null,"UserAccID":null,"AuthorKeywords":null,"OtherDescriptors":null,"Notes":null,"AnaPub":2021,"MonPub":null,"DateUpdate":"2021-12-08","DateCreate":"2021-12-08","SecASFANote":null,"ConfID":null,"PeerRev":1,"VlizCoreFlag":1,"WoScode":"WOS:000725094400043","VABBcode":null,"OpenAcc":1,"DOI":"10.1038/s41598-021-02387-9"},"refs":null,"anarec":{"AnaID":347969,"PubliDate":2021,"Pagination":"23209","XtraPublOfAnaID":null,"ISBN":null,"Volume":"11","Issue":null,"BRefMon":null,"BRefMonRR":null,"BRefXtra":null,"BRefXtraRR":null,"SerBRefID":208093,"SerRR":"Scientific Reports (Nature Publishing Group). Nature Publishing Group: London.  ISSN 2045-2322; e-ISSN 2045-2322","StandardTitleSer":"Scientific Reports (Nature Publishing Group)","ISSN":"2045-2322","AbbrevSer":"NPG Scientific Reports","StandardTitleMon":null,"StartPage":23209,"Pages":null,"ToPubliDate":null,"BRefBibLvlCode":"S","SerNotes":null},"monrec":null,"serrec":null,"relations":null,"relationsRev":null,"addrec":null,"othpubs":null,"ownerships":null,"authors":[{"AutName":"Hoekendijk","Firstname":"Jeroen","Initials":"J.P.A.","Affiliation":null,"Discriminator":null,"CorporateFlag":0,"BEntID":344661,"AutID":468064,"OrderNr":1,"DegrID":null,"EditorFlag":0,"CorrespFlag":0,"IllustratorFlag":0,"ReviserFlag":0,"TranslatorFlag":0,"InsAcronym":"COS","InsFSN":"Koninklijk Nederlands Instituut voor Onderzoek der Zee; Coastal Sciences","ORCID":"0000-0002-9308-7429","PersID":40568,"InsID":13656},{"AutName":"Kellenberger","Firstname":"Benjamin","Initials":"B.","Affiliation":null,"Discriminator":null,"CorporateFlag":0,"BEntID":344661,"AutID":472325,"OrderNr":2,"DegrID":null,"EditorFlag":0,"CorrespFlag":0,"IllustratorFlag":0,"ReviserFlag":0,"TranslatorFlag":0,"InsAcronym":null,"InsFSN":null,"ORCID":null,"PersID":null,"InsID":null},{"AutName":"Aarts","Firstname":"Geert","Initials":"G.M","Affiliation":"COS","Discriminator":null,"CorporateFlag":0,"BEntID":344661,"AutID":237715,"OrderNr":3,"DegrID":null,"EditorFlag":0,"CorrespFlag":0,"IllustratorFlag":0,"ReviserFlag":0,"TranslatorFlag":0,"InsAcronym":"COS","InsFSN":"Koninklijk Nederlands Instituut voor Onderzoek der Zee; Coastal Sciences","ORCID":"0000-0003-3196-4235","PersID":25190,"InsID":13656},{"AutName":"Brasseur","Firstname":"Sophie","Initials":"S.","Affiliation":null,"Discriminator":null,"CorporateFlag":0,"BEntID":344661,"AutID":218555,"OrderNr":4,"DegrID":null,"EditorFlag":0,"CorrespFlag":0,"IllustratorFlag":0,"ReviserFlag":0,"TranslatorFlag":0,"InsAcronym":null,"InsFSN":null,"ORCID":null,"PersID":null,"InsID":null},{"AutName":"Poiesz","Firstname":"Suzanne","Initials":"S.","Affiliation":"COS","Discriminator":null,"CorporateFlag":0,"BEntID":344661,"AutID":349947,"OrderNr":5,"DegrID":null,"EditorFlag":0,"CorrespFlag":0,"IllustratorFlag":0,"ReviserFlag":0,"TranslatorFlag":0,"InsAcronym":"COS","InsFSN":"Koninklijk Nederlands Instituut voor Onderzoek der Zee; Coastal Sciences","ORCID":"0000-0003-1396-3618","PersID":36724,"InsID":13656},{"AutName":"Tuia","Firstname":"Devis","Initials":"D.","Affiliation":null,"Discriminator":null,"CorporateFlag":0,"BEntID":344661,"AutID":472329,"OrderNr":6,"DegrID":null,"EditorFlag":0,"CorrespFlag":0,"IllustratorFlag":0,"ReviserFlag":0,"TranslatorFlag":0,"InsAcronym":null,"InsFSN":null,"ORCID":null,"PersID":null,"InsID":null}],"mapdetails":null,"datasets":null,"monographs":null,"monparts":null,"serparts":null,"BEntOpen":344661,"BEntPrivate":null,"availability":[{"BInstID":369906,"LibID":2779,"BRefID":347969,"EmbargoDate":null,"FullEmbargoDate":null,"PhysMedID":16,"hasOCRd":1,"ShelfLocCode":"369906","RFID":null,"PaidValue":null,"Medium":"Server","Description":null,"Acronym":null,"Library":"NIOZ","DutchTerm":null,"URL":null,"ClassifID":260,"Classification":"NIOZ Open Repository","ReqLink":null,"ClassifTypID":1,"URLLocation":"https://www.vliz.be/imisdocs/publications/","SubDir":1,"InternalReq":null,"LoggedInReq":null,"Disclaimer":"Disclaimer_NIOZ","DutchDisclaimer":null,"FileFormat":".pdf","FileDescr":"pdf","InsPub":1,"InsID":397,"FileFormID":6,"LendableFlag":null,"PublicFlag":1,"orderLib":"NIOZ","Notes":null,"AccConID":null,"AccessConstraint":null,"LicURL":null}],"litstyles":null,"thespers":null,"arch2discl":805,"SERpubls":[{"PublName":"Nature Publishing Group","City":"London"}],"MONpubls":null,"pictures":[],"thestermsPath":null,"thestermsASFA":null,"taxtermsASFA":null,"geotermsASFA":null,"collections":null,"conf":null,"proj":null,"Physdatasets":null,"spcols":{"805":{"SpName":"Koninklijk Nederlands Instituut voor Onderzoek der Zee","SpColID":805,"ParSpColID":null,"TopParID":null,"ShortName":"NIOZ","URLLocation":"https://www.vliz.be/imis/nioz/imis.php?refid=","LibID":2779,"OpenRepoFlag":1,"SpTypID":1,"TopParIDNotWebsite":null,"SpColPath":"NIOZ"}},"doi":null,"publs":null,"serparttypes":null,"monauthors":null,"MParts":null,"SParts":null,"hLibs":null,"langs":[{"BEntID":344661,"AbstractFlag":0,"LangID":15,"LangCode":"en","Lang":"English","DutchTerm":"Engels","LangCodeExtended":"eng"},{"BEntID":344661,"AbstractFlag":1,"LangID":15,"LangCode":"en","Lang":"English","DutchTerm":"Engels","LangCodeExtended":"eng"}],"urls":[{"URL":"https://doi.org/10.25850/nioz/7b.b0c","externalID":"10.25850/nioz/7b.b0c","URLTypeCode":null,"URLID":98314,"URLTypID":37,"URLType":"Additional info","URLPrefix":null},{"URL":"https://dx.doi.org/10.1038/s41598-021-02387-9","externalID":"10.1038/s41598-021-02387-9","URLTypeCode":"DOI","URLID":98313,"URLTypID":13,"URLType":"DOI","URLPrefix":"http://dx.doi.org/"}],"thesterms":null,"taxterms":null,"geoterms":null,"othterms":null,"asfacodes":null,"asfa2codes":null,"thestermsFRIS":null,"taxtermsFRIS":null,"geotermsFRIS":null,"othtermsFRIS":null,"resmessage":"","complete":1,"sessions":{"newSesName":"Leonne.van.der.Weegen@nioz.nl","newSesDate":{"date":"2021-12-08 10:46:21.597000","timezone_type":3,"timezone":"Europe/Brussels"},"updSesName":"Leonne.van.der.Weegen@nioz.nl","updSesDate":{"date":"2021-12-08 10:46:21.597000","timezone_type":3,"timezone":"Europe/Brussels"}}}
