Remoção de objetos em imagens do mercado imobiliário usando inpainting
Autor(a) principal: | |
---|---|
Data de Publicação: | 2023 |
Tipo de documento: | Dissertação |
Idioma: | eng |
Título da fonte: | Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
Texto Completo: | http://hdl.handle.net/10400.22/23868 |
Resumo: | The real estate market is one of the most valuable and influential markets in the world. An essential part of this market is the first contact with potential customers through property listings on real estate websites. To make listings more attractive, professionals often hire external services or use image editing software to improve the quality of the images shown. Real estate agents often use tools to remove unwanted objects from images to declutter rooms, draw attention to the property, and eliminate uncontrollable environmental elements. However, existing solutions are not cost-effective for real estate agencies, as they have to pay for each image and the results can take up to two days to be delivered. In recent years, the use of deep learning artificial intelligence algorithms has revolutionised image removal technology, with current technologies capable of producing natural and realistic results. This paper focusses on the development of a deep learning inpainting image object removal solution for real estate images to be integrated into Maxwork, the back-office portal used by RE/MAX Portugal, one of the largest real estate companies in Portugal. This solution can remove objects from real estate images and contains several additional features, including the ability to undo and redo, and compare the original image with the most recent result. It uses the LaMa inpainting deep learning model, which proved to be more effective than other state-of-the-art models. The effectiveness of the solution was evaluated with several objective metrics, such as Fréchet Inception Distance (FID) and Learned Perceptual Image Patch Similarity (LPIPS), and subjectively with user feedback that gave an average rating of 4.31 for ease of use and 3.90 for satisfaction with the results obtained. Furthermore, ethical considerations related to image editing in the real estate sector are discussed to ensure a transparent and honest use of the solution. This paper also outlines the details of an experiment to train a new deep learning inpainting model using a collection of around 100,000 real estate images from past sold listings. However, the results of the experiment showed that the trained model was not able to outperform the pre-trained LaMa model, as it scored worse across all metrics. Additionally, this paper provides research on existing real estate image object removal solutions, the current application fields of deep learning inpainting models, and the evolution of inpainting models from traditional methods to current state-of-the-art deep learning models. |
id |
RCAP_e56f3e53d55c6673e842497adbe9fdb5 |
---|---|
oai_identifier_str |
oai:recipp.ipp.pt:10400.22/23868 |
network_acronym_str |
RCAP |
network_name_str |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
repository_id_str |
7160 |
spelling |
Remoção de objetos em imagens do mercado imobiliário usando inpaintingReal EstateInpaintingImage Object RemovalComputer VisionLaMaImobiliárioRemoção Objetos ImagemVisão ComputacionalThe real estate market is one of the most valuable and influential markets in the world. An essential part of this market is the first contact with potential customers through property listings on real estate websites. To make listings more attractive, professionals often hire external services or use image editing software to improve the quality of the images shown. Real estate agents often use tools to remove unwanted objects from images to declutter rooms, draw attention to the property, and eliminate uncontrollable environmental elements. However, existing solutions are not cost-effective for real estate agencies, as they have to pay for each image and the results can take up to two days to be delivered. In recent years, the use of deep learning artificial intelligence algorithms has revolutionised image removal technology, with current technologies capable of producing natural and realistic results. This paper focusses on the development of a deep learning inpainting image object removal solution for real estate images to be integrated into Maxwork, the back-office portal used by RE/MAX Portugal, one of the largest real estate companies in Portugal. This solution can remove objects from real estate images and contains several additional features, including the ability to undo and redo, and compare the original image with the most recent result. It uses the LaMa inpainting deep learning model, which proved to be more effective than other state-of-the-art models. The effectiveness of the solution was evaluated with several objective metrics, such as Fréchet Inception Distance (FID) and Learned Perceptual Image Patch Similarity (LPIPS), and subjectively with user feedback that gave an average rating of 4.31 for ease of use and 3.90 for satisfaction with the results obtained. Furthermore, ethical considerations related to image editing in the real estate sector are discussed to ensure a transparent and honest use of the solution. This paper also outlines the details of an experiment to train a new deep learning inpainting model using a collection of around 100,000 real estate images from past sold listings. However, the results of the experiment showed that the trained model was not able to outperform the pre-trained LaMa model, as it scored worse across all metrics. Additionally, this paper provides research on existing real estate image object removal solutions, the current application fields of deep learning inpainting models, and the evolution of inpainting models from traditional methods to current state-of-the-art deep learning models.O mercado imobiliário é um dos mercados mais valiosos e influentes do mundo. Uma parte essencial desse mercado é o primeiro contato com potenciais clientes por meio de anúncios de imóveis em sites imobiliários. Para tornar estes anúncios mais atrativos, os profissionais muitas vezes contratam serviços externos ou usam software de edição de imagens para melhorar a qualidade das imagens exibidas. Agentes imobiliários frequentemente usam ferramentas para remover objetos indesejados das imagens, a fim de desobstruir os quartos, chamar a atenção para a propriedade e eliminar elementos ambientais incontroláveis. No entanto, as soluções existentes não são economicamente viáveis para agências imobiliárias, pois elas têm de pagar por cada imagem e os resultados podem levar até dois dias para serem entregues. Nos últimos anos, o uso de algoritmos de inteligência artificial de aprendizagem profunda revolucionou a tecnologia de remoção de imagens, com as tecnologias atuais capazes de produzir resultados naturais e realistas. Este trabalho foca-se no desenvolvimento de uma solução de remoção de objetos de imagem usando aprendizagem profunda para imagens imobiliárias, a ser integrada no Maxwork, o portal de backoffice usado pela RE/MAX Portugal, uma das maiores empresas imobiliárias em Portugal. Esta solução é capaz de remover objetos de imagens imobiliárias e contém diversas funcionalidades adicionais, incluindo a capacidade de desfazer e refazer ações, bem como comparar a imagem original com o resultado mais recente. Ela utiliza o modelo de aprendizagem profunda LaMa, que se mostrou mais eficaz do que outros modelos estado da arte. A eficácia da solução foi avaliada por meio de várias métricas objetivas, como Fréchet Inception Distance (FID) e Learned Perceptual Image Patch Similarity (LPIPS), e subjetivamente com feedback de um grupo de utilizadores, que deram uma classificação média de 4.31 para a facilidade de uso e 3.90 para satisfação com os resultados obtidos. Além disso, são discutidas considerações éticas relacionadas à edição de imagens no setor imobiliário, para garantir um uso transparente e honesto da solução. Este trabalho também descreve os detalhes de uma experiência para treinar um novo modelo de remoção de objetos de imagens usando uma coleção de cerca de 100.000 imagens imobiliárias de anúncios passados. No entanto, os resultados da experiência mostraram que o modelo treinado não conseguiu superar o modelo LaMa pré-treinado, pois obteve resultados piores em todas as métricas. Além disso, este trabalho oferece uma pesquisa sobre soluções existentes de remoção de objetos de imagens imobiliárias, e os campos de aplicação atuais e a evolução deste tipo de modelos, desde métodos tradicionais até modelos estado da arte de aprendizagem profunda.Ramos, Carlos Fernando da SilvaRepositório Científico do Instituto Politécnico do PortoReis, André Ferreira Alves dos2023-11-08T15:58:29Z20232023-01-01T00:00:00Zinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisapplication/pdfhttp://hdl.handle.net/10400.22/23868TID:203380231enginfo:eu-repo/semantics/openAccessreponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãoinstacron:RCAAP2023-11-15T01:47:46Zoai:recipp.ipp.pt:10400.22/23868Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireopendoar:71602024-03-19T22:42:33.637957Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãofalse |
dc.title.none.fl_str_mv |
Remoção de objetos em imagens do mercado imobiliário usando inpainting |
title |
Remoção de objetos em imagens do mercado imobiliário usando inpainting |
spellingShingle |
Remoção de objetos em imagens do mercado imobiliário usando inpainting Reis, André Ferreira Alves dos Real Estate Inpainting Image Object Removal Computer Vision LaMa Imobiliário Remoção Objetos Imagem Visão Computacional |
title_short |
Remoção de objetos em imagens do mercado imobiliário usando inpainting |
title_full |
Remoção de objetos em imagens do mercado imobiliário usando inpainting |
title_fullStr |
Remoção de objetos em imagens do mercado imobiliário usando inpainting |
title_full_unstemmed |
Remoção de objetos em imagens do mercado imobiliário usando inpainting |
title_sort |
Remoção de objetos em imagens do mercado imobiliário usando inpainting |
author |
Reis, André Ferreira Alves dos |
author_facet |
Reis, André Ferreira Alves dos |
author_role |
author |
dc.contributor.none.fl_str_mv |
Ramos, Carlos Fernando da Silva Repositório Científico do Instituto Politécnico do Porto |
dc.contributor.author.fl_str_mv |
Reis, André Ferreira Alves dos |
dc.subject.por.fl_str_mv |
Real Estate Inpainting Image Object Removal Computer Vision LaMa Imobiliário Remoção Objetos Imagem Visão Computacional |
topic |
Real Estate Inpainting Image Object Removal Computer Vision LaMa Imobiliário Remoção Objetos Imagem Visão Computacional |
description |
The real estate market is one of the most valuable and influential markets in the world. An essential part of this market is the first contact with potential customers through property listings on real estate websites. To make listings more attractive, professionals often hire external services or use image editing software to improve the quality of the images shown. Real estate agents often use tools to remove unwanted objects from images to declutter rooms, draw attention to the property, and eliminate uncontrollable environmental elements. However, existing solutions are not cost-effective for real estate agencies, as they have to pay for each image and the results can take up to two days to be delivered. In recent years, the use of deep learning artificial intelligence algorithms has revolutionised image removal technology, with current technologies capable of producing natural and realistic results. This paper focusses on the development of a deep learning inpainting image object removal solution for real estate images to be integrated into Maxwork, the back-office portal used by RE/MAX Portugal, one of the largest real estate companies in Portugal. This solution can remove objects from real estate images and contains several additional features, including the ability to undo and redo, and compare the original image with the most recent result. It uses the LaMa inpainting deep learning model, which proved to be more effective than other state-of-the-art models. The effectiveness of the solution was evaluated with several objective metrics, such as Fréchet Inception Distance (FID) and Learned Perceptual Image Patch Similarity (LPIPS), and subjectively with user feedback that gave an average rating of 4.31 for ease of use and 3.90 for satisfaction with the results obtained. Furthermore, ethical considerations related to image editing in the real estate sector are discussed to ensure a transparent and honest use of the solution. This paper also outlines the details of an experiment to train a new deep learning inpainting model using a collection of around 100,000 real estate images from past sold listings. However, the results of the experiment showed that the trained model was not able to outperform the pre-trained LaMa model, as it scored worse across all metrics. Additionally, this paper provides research on existing real estate image object removal solutions, the current application fields of deep learning inpainting models, and the evolution of inpainting models from traditional methods to current state-of-the-art deep learning models. |
publishDate |
2023 |
dc.date.none.fl_str_mv |
2023-11-08T15:58:29Z 2023 2023-01-01T00:00:00Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/masterThesis |
format |
masterThesis |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://hdl.handle.net/10400.22/23868 TID:203380231 |
url |
http://hdl.handle.net/10400.22/23868 |
identifier_str_mv |
TID:203380231 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
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.source.none.fl_str_mv |
reponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação instacron:RCAAP |
instname_str |
Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação |
instacron_str |
RCAAP |
institution |
RCAAP |
reponame_str |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
collection |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
repository.name.fl_str_mv |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação |
repository.mail.fl_str_mv |
|
_version_ |
1799134990803402752 |