AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO

Detalhes bibliográficos
Autor(a) principal: Lucas Prado Osco
Data de Publicação: 2023
Tipo de documento: Tese
Idioma: por
Título da fonte: Repositório Institucional da UFMS
Texto Completo: https://repositorio.ufms.br/handle/123456789/6720
Resumo: This thesis consists of an analysis of recent innovations in deep learning (DL) techniques, applied to remote sensing images, with a focus on advancements in Deep Neural Networks (DNN), Visual Language Models (VLM), and zero-shot segmentation with the Segment Anything Model (SAM). The contribution of this work lies in providing a discussion of the state of the art of these technologies within the context of information extraction from remote sensing images. Drawing on literature reviews, model analyses and adaptations, and experiments with remote sensing datasets, the thesis is organized into chapters. The first chapter offers a literature review of the application of DNNs to high spatial resolution images, obtained by sensors onboard Unmanned Aerial Vehicles (UAVs). Here, we analyze 232 scientific articles and demonstrate that DL shows promising results for a range of applications concerning aerial image processing tasks. The second chapter explores the application of Visual ChatGPT, an innovation in VLM, within the remote sensing context. Despite being in the early stages of development, Visual ChatGPT, with its ability to analyze images based on textual inputs, could revolutionize the digital processing of remote sensing images, creating opportunities and optimizing the information extraction process. The third and final chapter examines the performance of SAM in segmenting remote sensing images across multiple scales, representative of varied and challenging geographical contexts. Despite its limitations in images with metric resolution, SAM demonstrates satisfactory performance in segmentation when compared to human manual annotation in multiple cases. In summary, this thesis compiles the latest in the context of applying DL models to multiscale remote sensing images. It establishes both the advancements and challenges to be overcome in this field, outlining paths for future research aimed at assessing remote sensing images in various applications.
id UFMS_4c3d3420a363aba3676307cc0dba605a
oai_identifier_str oai:repositorio.ufms.br:123456789/6720
network_acronym_str UFMS
network_name_str Repositório Institucional da UFMS
repository_id_str 2124
spelling 2023-10-30T12:40:37Z2023-10-30T12:40:37Z2023https://repositorio.ufms.br/handle/123456789/6720This thesis consists of an analysis of recent innovations in deep learning (DL) techniques, applied to remote sensing images, with a focus on advancements in Deep Neural Networks (DNN), Visual Language Models (VLM), and zero-shot segmentation with the Segment Anything Model (SAM). The contribution of this work lies in providing a discussion of the state of the art of these technologies within the context of information extraction from remote sensing images. Drawing on literature reviews, model analyses and adaptations, and experiments with remote sensing datasets, the thesis is organized into chapters. The first chapter offers a literature review of the application of DNNs to high spatial resolution images, obtained by sensors onboard Unmanned Aerial Vehicles (UAVs). Here, we analyze 232 scientific articles and demonstrate that DL shows promising results for a range of applications concerning aerial image processing tasks. The second chapter explores the application of Visual ChatGPT, an innovation in VLM, within the remote sensing context. Despite being in the early stages of development, Visual ChatGPT, with its ability to analyze images based on textual inputs, could revolutionize the digital processing of remote sensing images, creating opportunities and optimizing the information extraction process. The third and final chapter examines the performance of SAM in segmenting remote sensing images across multiple scales, representative of varied and challenging geographical contexts. Despite its limitations in images with metric resolution, SAM demonstrates satisfactory performance in segmentation when compared to human manual annotation in multiple cases. In summary, this thesis compiles the latest in the context of applying DL models to multiscale remote sensing images. It establishes both the advancements and challenges to be overcome in this field, outlining paths for future research aimed at assessing remote sensing images in various applications.Esta tese consiste-se em uma análise das recentes inovações em técnicas de aprendizagem profunda (Deep Learning - DL), aplicadas em imagens de sensoriamento remoto, com foco nos avanços em Redes Neurais Profundas (Deep Neural Networks – DNN), Modelos de Linguagem Visual (Visual Language Models - VLM) e em segmentação zero-shot com o Segment Anything Model (SAM). A contribuição deste trabalho está em fornecer uma discussão do estado da arte dessas tecnologias no contexto da extração de informações em imagens de sensoriamento remoto. Baseando-se em revisões de literatura, análises e adaptações de modelos e experimentos com conjuntos de dados de sensoriamento remoto, a tese encontra-se organizada em forma de capítulos. O primeiro capítulo oferece uma revisão da literatura da aplicação de DNNs em imagens de alta resolução espacial, adquiridas por sensores embarcados em Veículos Aéreos Não-Tripulados (VANTs). Aqui, analisamos 232 artigos científicos e demonstramos que DL apresenta resultados promissores para uma série de aplicações no que se referem às tarefas de processamento de imagens aéreas. O segundo capítulo explora a aplicação do Visual ChatGPT, uma inovação em VLM, no contexto do sensoriamento remoto. Apesar de estar em fase inicial de desenvolvimento, o Visual ChatGPT, com as suas habilidades baseadas em entradas textuais para analisar imagens, poderá revolucionar o processamento digital de imagens de sensoriamento remoto, criando oportunidades e otimizando o processo de extração da informação. O terceiro e último capítulo examina o desempenho do SAM em segmentar imagens de sensoriamento remoto de múltiplas escalas, representativas de variados e desafiadores contextos geográficos. Apesar de suas limitações em imagens com resolução métrica, SAM demonstra desempenho satisfatório na segmentação quando equiparado a anotação manual humana em múltiplos casos. Em suma, esta tese compila o que há de mais recente no contexto de aplicação de modelos de DL em imagens de sensoriamento remoto de multiescala. Constata-se aqui, tanto os avanços quanto os desafios a serem superados neste campo, delineando caminhos para pesquisas futuras que visem avaliar imagens de sensoriamento remoto em aplicações diversas.Fundação Universidade Federal de Mato Grosso do SulUFMSBrasilSensoriamento remotoAVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTOinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/doctoralThesisJose Marcato JuniorLucas Prado Oscoinfo:eu-repo/semantics/openAccessporreponame:Repositório Institucional da UFMSinstname:Universidade Federal de Mato Grosso do Sul (UFMS)instacron:UFMSORIGINALTeseUFMS_LucasOsco_v4_Combinado_Comp.pdfTeseUFMS_LucasOsco_v4_Combinado_Comp.pdfapplication/pdf4494063https://repositorio.ufms.br/bitstream/123456789/6720/-1/TeseUFMS_LucasOsco_v4_Combinado_Comp.pdfb7cff0cc599f27c8213753158c9c1eadMD5-1123456789/67202023-10-30 08:40:38.221oai:repositorio.ufms.br:123456789/6720Repositório InstitucionalPUBhttps://repositorio.ufms.br/oai/requestri.prograd@ufms.bropendoar:21242023-10-30T12:40:38Repositório Institucional da UFMS - Universidade Federal de Mato Grosso do Sul (UFMS)false
dc.title.pt_BR.fl_str_mv AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO
title AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO
spellingShingle AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO
Lucas Prado Osco
Sensoriamento remoto
title_short AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO
title_full AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO
title_fullStr AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO
title_full_unstemmed AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO
title_sort AVANÇOS EM APRENDIZAGEM PROFUNDA APLICADA AO SENSORIAMENTO REMOTO
author Lucas Prado Osco
author_facet Lucas Prado Osco
author_role author
dc.contributor.advisor1.fl_str_mv Jose Marcato Junior
dc.contributor.author.fl_str_mv Lucas Prado Osco
contributor_str_mv Jose Marcato Junior
dc.subject.por.fl_str_mv Sensoriamento remoto
topic Sensoriamento remoto
description This thesis consists of an analysis of recent innovations in deep learning (DL) techniques, applied to remote sensing images, with a focus on advancements in Deep Neural Networks (DNN), Visual Language Models (VLM), and zero-shot segmentation with the Segment Anything Model (SAM). The contribution of this work lies in providing a discussion of the state of the art of these technologies within the context of information extraction from remote sensing images. Drawing on literature reviews, model analyses and adaptations, and experiments with remote sensing datasets, the thesis is organized into chapters. The first chapter offers a literature review of the application of DNNs to high spatial resolution images, obtained by sensors onboard Unmanned Aerial Vehicles (UAVs). Here, we analyze 232 scientific articles and demonstrate that DL shows promising results for a range of applications concerning aerial image processing tasks. The second chapter explores the application of Visual ChatGPT, an innovation in VLM, within the remote sensing context. Despite being in the early stages of development, Visual ChatGPT, with its ability to analyze images based on textual inputs, could revolutionize the digital processing of remote sensing images, creating opportunities and optimizing the information extraction process. The third and final chapter examines the performance of SAM in segmenting remote sensing images across multiple scales, representative of varied and challenging geographical contexts. Despite its limitations in images with metric resolution, SAM demonstrates satisfactory performance in segmentation when compared to human manual annotation in multiple cases. In summary, this thesis compiles the latest in the context of applying DL models to multiscale remote sensing images. It establishes both the advancements and challenges to be overcome in this field, outlining paths for future research aimed at assessing remote sensing images in various applications.
publishDate 2023
dc.date.accessioned.fl_str_mv 2023-10-30T12:40:37Z
dc.date.available.fl_str_mv 2023-10-30T12:40:37Z
dc.date.issued.fl_str_mv 2023
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.driver.fl_str_mv info:eu-repo/semantics/doctoralThesis
format doctoralThesis
status_str publishedVersion
dc.identifier.uri.fl_str_mv https://repositorio.ufms.br/handle/123456789/6720
url https://repositorio.ufms.br/handle/123456789/6720
dc.language.iso.fl_str_mv por
language por
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.publisher.none.fl_str_mv Fundação Universidade Federal de Mato Grosso do Sul
dc.publisher.initials.fl_str_mv UFMS
dc.publisher.country.fl_str_mv Brasil
publisher.none.fl_str_mv Fundação Universidade Federal de Mato Grosso do Sul
dc.source.none.fl_str_mv reponame:Repositório Institucional da UFMS
instname:Universidade Federal de Mato Grosso do Sul (UFMS)
instacron:UFMS
instname_str Universidade Federal de Mato Grosso do Sul (UFMS)
instacron_str UFMS
institution UFMS
reponame_str Repositório Institucional da UFMS
collection Repositório Institucional da UFMS
bitstream.url.fl_str_mv https://repositorio.ufms.br/bitstream/123456789/6720/-1/TeseUFMS_LucasOsco_v4_Combinado_Comp.pdf
bitstream.checksum.fl_str_mv b7cff0cc599f27c8213753158c9c1ead
bitstream.checksumAlgorithm.fl_str_mv MD5
repository.name.fl_str_mv Repositório Institucional da UFMS - Universidade Federal de Mato Grosso do Sul (UFMS)
repository.mail.fl_str_mv ri.prograd@ufms.br
_version_ 1801678533795250176