Development of a new non-invasive vineyard yield estimation method based on image analysis

Bibliographic Details
Main Author: Victorino, Gonçalo Filipe dos Santos
Publication Date: 2022
Language: eng
Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
Download full: http://hdl.handle.net/10400.5/27513
Summary: Doutoramento em Engenharia Agronómica / Instituto Superior de Agronomia. Universidade de Lisboa
id RCAP_e404252dca2a91a9f2a8b7924a42d575
oai_identifier_str oai:repositorio.ulisboa.pt:10400.5/27513
network_acronym_str RCAP
network_name_str Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
repository_id_str https://opendoar.ac.uk/repository/7160
spelling Development of a new non-invasive vineyard yield estimation method based on image analysisgrapevine yield predictionbunch occlusionimage analysisnon-invasivebunch featuresregression modelDoutoramento em Engenharia Agronómica / Instituto Superior de Agronomia. Universidade de LisboaPredicting vineyard yield with accuracy can provide several advantages to the whole vine and wine industry. Today this is majorly done using manual and sometimes destructive methods, based on bunch samples. Yield estimation using computer vision and image analysis can potentially perform this task extensively, automatically, and non-invasively. In the present work this approach is explored in three main steps: image collection, occluded fruit estimation and image traits conversion to mass. On the first step, grapevine images were collected in field conditions along some of the main grapevine phenological stages. Visible yield components were identified in the image and compared to ground truth. When analyzing inflorescences and bunches, more than 50% were occluded by leaves or other plant organs, on three cultivars. No significant differences were observed on bunch visibility after fruit set. Visible bunch projected area explained an average of 49% of vine yield variation, between veraison and harvest. On the second step, vine images were collected, in field conditions, with different levels of defoliation intensity at bunch zone. A regression model was computed combining canopy porosity and visible bunch area, obtained via image analysis, which explained 70-84% of bunch exposure variation. This approach allowed for an estimation of the occluded fraction of bunches with average errors below |10|%. No significant differences were found between the model’s output at veraison and harvest. On the last step, the conversion of bunch image traits into mass was explored in laboratory and field conditions. In both cases, cultivar differences related to bunch architecture were found to affect weight estimation. A combination of derived variables which included visible bunch area, estimated total bunch area, visible bunch perimeter, visible berry number and bunch compactness was used to estimate yield on undisturbed grapevines. The final model achieved a R2 = 0.86 between actual and estimated yield (n = 213). If performed automatically, the final approach suggested in this work has the potential to provide a non-invasive method that can be performed accurately across whole vineyards.ISA/ULLopes, Carlos Manuel AntunesBraga, Ricardo Nuno da Fonseca Garcia PereiraSantos-Victor, José AlbertoRepositório da Universidade de LisboaVictorino, Gonçalo Filipe dos Santos2023-03-27T10:48:26Z20222022-01-01T00:00:00Zdoctoral thesisinfo:eu-repo/semantics/publishedVersionapplication/pdfhttp://hdl.handle.net/10400.5/27513TID:101713665engVictorino, G.F.S. - Development of a new non-invasive vineyard yield estimation method based on image analysis. Lisboa: ISA, 2022, 130 p.info:eu-repo/semantics/openAccessreponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiainstacron:RCAAP2025-03-17T16:10:18Zoai:repositorio.ulisboa.pt:10400.5/27513Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireinfo@rcaap.ptopendoar:https://opendoar.ac.uk/repository/71602025-05-29T04:04:50.378969Repositórios Científicos de Acesso Aberto de Portugal (RCAAP) - FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiafalse
dc.title.none.fl_str_mv Development of a new non-invasive vineyard yield estimation method based on image analysis
title Development of a new non-invasive vineyard yield estimation method based on image analysis
spellingShingle Development of a new non-invasive vineyard yield estimation method based on image analysis
Victorino, Gonçalo Filipe dos Santos
grapevine yield prediction
bunch occlusion
image analysis
non-invasive
bunch features
regression model
title_short Development of a new non-invasive vineyard yield estimation method based on image analysis
title_full Development of a new non-invasive vineyard yield estimation method based on image analysis
title_fullStr Development of a new non-invasive vineyard yield estimation method based on image analysis
title_full_unstemmed Development of a new non-invasive vineyard yield estimation method based on image analysis
title_sort Development of a new non-invasive vineyard yield estimation method based on image analysis
author Victorino, Gonçalo Filipe dos Santos
author_facet Victorino, Gonçalo Filipe dos Santos
author_role author
dc.contributor.none.fl_str_mv Lopes, Carlos Manuel Antunes
Braga, Ricardo Nuno da Fonseca Garcia Pereira
Santos-Victor, José Alberto
Repositório da Universidade de Lisboa
dc.contributor.author.fl_str_mv Victorino, Gonçalo Filipe dos Santos
dc.subject.por.fl_str_mv grapevine yield prediction
bunch occlusion
image analysis
non-invasive
bunch features
regression model
topic grapevine yield prediction
bunch occlusion
image analysis
non-invasive
bunch features
regression model
description Doutoramento em Engenharia Agronómica / Instituto Superior de Agronomia. Universidade de Lisboa
publishDate 2022
dc.date.none.fl_str_mv 2022
2022-01-01T00:00:00Z
2023-03-27T10:48:26Z
dc.type.driver.fl_str_mv doctoral thesis
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
status_str publishedVersion
dc.identifier.uri.fl_str_mv http://hdl.handle.net/10400.5/27513
TID:101713665
url http://hdl.handle.net/10400.5/27513
identifier_str_mv TID:101713665
dc.language.iso.fl_str_mv eng
language eng
dc.relation.none.fl_str_mv Victorino, G.F.S. - Development of a new non-invasive vineyard yield estimation method based on image analysis. Lisboa: ISA, 2022, 130 p.
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.publisher.none.fl_str_mv ISA/UL
publisher.none.fl_str_mv ISA/UL
dc.source.none.fl_str_mv reponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
instacron:RCAAP
instname_str FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
instacron_str RCAAP
institution RCAAP
reponame_str Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
collection Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
repository.name.fl_str_mv Repositórios Científicos de Acesso Aberto de Portugal (RCAAP) - FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
repository.mail.fl_str_mv info@rcaap.pt
_version_ 1833601934917894144

Similar Items