Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images
Autor(a) principal: | |
---|---|
Data de Publicação: | 2019 |
Tipo de documento: | Dissertação |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UFMG |
Texto Completo: | http://hdl.handle.net/1843/36706 |
Resumo: | High-resolution aerial images are desirable for most of the deep-based remote sensing applications. This type of data, however, is not always accessible or affordable. On the other hand, coarse resolution remote sensing images, such as LANDSAT and MODIS, are easily found in public open repositories and, therefore, are widely used in many studies. The problem is that the amount of spacial information compressed into one single pixel in a low-resolution representation can compromise pattern recognition algorithms. Thus, the use of coarse-resolution data for automatic creation of thematic maps is very restricted since most of the deep-based semantic segmentation (a.k.a dense labeling) approaches are only suitable for subdecimeter data. Super-resolution is a classic computer vision problem that aims to restore the quality of degraded, low-resolution images. In this work, we design two frameworks in order to evaluate the effectiveness of deep-based super-resolution in the semantic segmentation of low-resolution remote sensing images. Our objective is to evaluate how effective is deep-based super-resolution to different levels of degradation, how it compares to unsupervised bicubic interpolation and if it is able to reconstruct small objects and, consequently, contribute to semantic segmentation improvement. The first framework uses super-resolution as a pre-processing step for the semantic segmentation task (two-stage framework). The second framework is an end-to-end approach that trains both networks at the same time while sharing their losses. We carried out an extensive set of experiments on remote sensing datasets with distinct nature and properties. For the agricultural dataset of coffee mapping, which only contains two labels (coffee and non-coffee), the use of low-resolution images achieved only 50% normalized accuracy with 8x up-scaling factor. The two stage framework with super-resolution in the same condition increased this value to 72%. The end-to-end framework further increased the value to 77%, compared to 81% of high-resolution data. For the urban dataset of Vaihingen, using super-resolution in the two stage framework increased the accuracy of car segmentation from 19% to 58% with 8x up-scaling factor, while the end-to-end framework achieved 65%. In this case, with high-resolution data, the accuracy was 69%, which is not far from the super-resolution result. Both cases are examples of how super-resolution is able to recover important texture details (for coffee crops, for example) and is also able to make more discernible small objects that were difficult to see in a low-resolution representation (such as cars). The results show that super-resolution is effective to improve semantic segmentation performance on low-resolution aerial imagery. It not only outperforms unsupervised interpolation but also achieves semantic segmentation results comparable to high-resolution data. Even with a few training data, the use of the frameworks still achieved better results than bicubic interpolation. Thus, using super-resolution has proven to be a more effective approach than directly inputting low-resolution images to a semantic segmentation network. This is especially true for high degrading factors, which are the cases that super-resolution surpasses more the performance of low-resolution data. |
id |
UFMG_abe5d33428e40c2c64eb4c3e903ee0f6 |
---|---|
oai_identifier_str |
oai:repositorio.ufmg.br:1843/36706 |
network_acronym_str |
UFMG |
network_name_str |
Repositório Institucional da UFMG |
repository_id_str |
|
spelling |
Jefersson Alex dos Santoshttp://lattes.cnpq.br/2171782600728348George Luiz Medeiros TeodoroAndré Vital SaúdeWesley Nunes Gonçalveshttp://lattes.cnpq.br/5014941996905154Matheus Barros Pereira2021-07-09T15:03:16Z2021-07-09T15:03:16Z2019-11-11http://hdl.handle.net/1843/36706High-resolution aerial images are desirable for most of the deep-based remote sensing applications. This type of data, however, is not always accessible or affordable. On the other hand, coarse resolution remote sensing images, such as LANDSAT and MODIS, are easily found in public open repositories and, therefore, are widely used in many studies. The problem is that the amount of spacial information compressed into one single pixel in a low-resolution representation can compromise pattern recognition algorithms. Thus, the use of coarse-resolution data for automatic creation of thematic maps is very restricted since most of the deep-based semantic segmentation (a.k.a dense labeling) approaches are only suitable for subdecimeter data. Super-resolution is a classic computer vision problem that aims to restore the quality of degraded, low-resolution images. In this work, we design two frameworks in order to evaluate the effectiveness of deep-based super-resolution in the semantic segmentation of low-resolution remote sensing images. Our objective is to evaluate how effective is deep-based super-resolution to different levels of degradation, how it compares to unsupervised bicubic interpolation and if it is able to reconstruct small objects and, consequently, contribute to semantic segmentation improvement. The first framework uses super-resolution as a pre-processing step for the semantic segmentation task (two-stage framework). The second framework is an end-to-end approach that trains both networks at the same time while sharing their losses. We carried out an extensive set of experiments on remote sensing datasets with distinct nature and properties. For the agricultural dataset of coffee mapping, which only contains two labels (coffee and non-coffee), the use of low-resolution images achieved only 50% normalized accuracy with 8x up-scaling factor. The two stage framework with super-resolution in the same condition increased this value to 72%. The end-to-end framework further increased the value to 77%, compared to 81% of high-resolution data. For the urban dataset of Vaihingen, using super-resolution in the two stage framework increased the accuracy of car segmentation from 19% to 58% with 8x up-scaling factor, while the end-to-end framework achieved 65%. In this case, with high-resolution data, the accuracy was 69%, which is not far from the super-resolution result. Both cases are examples of how super-resolution is able to recover important texture details (for coffee crops, for example) and is also able to make more discernible small objects that were difficult to see in a low-resolution representation (such as cars). The results show that super-resolution is effective to improve semantic segmentation performance on low-resolution aerial imagery. It not only outperforms unsupervised interpolation but also achieves semantic segmentation results comparable to high-resolution data. Even with a few training data, the use of the frameworks still achieved better results than bicubic interpolation. Thus, using super-resolution has proven to be a more effective approach than directly inputting low-resolution images to a semantic segmentation network. This is especially true for high degrading factors, which are the cases that super-resolution surpasses more the performance of low-resolution data.Imagens aéreas de alta resolução são desejáveis para a maior parte das aplicações de sensoriamento remoto baseadas em algoritmos profundos. Esse tipo de dado, contudo, nem sempre é acessível. Por outro lado, imagens de sensoriamento remoto de baixa/média resolução, como as dos satélites LANDSAT e MODIS, são facilmente encontradas em repositórios públicos abertos e, portanto, são usadas em diversos estudos. O problema é que a quantidade de informação espacial comprimida em um único pixel em uma representação de baixa resolução pode comprometer algoritmos de reconhecimento de padrão. Assim, o uso de dados de baixa resolução para a criação automática de mapas temáticos é muito restrito, dado que a maioria das abordagens baseadas em algoritmos profundos para segmentação semântica (ou rotulação densa) são adequadas apenas para dados subdecimais. Super-resolução é um problema clássico de visão computacional que busca restaurar a qualidade de imagens de baixa resolução. No presente trabalho, foram desenvolvidos dois arcabouços que têm como objetivo avaliar a efetividade de super-resolução baseada em algoritmos profundos na segmentação semântica de imagens de sensoriamento remoto de baixa resolução. Visa-se avaliar quão efetivo é a super-resolução em diferentes níveis de degradação, como se compara com interpolação bicúbica não-supervisionada e se é capaz de reconstruir objetos pequenos e, consequentemente, contribuir para o melhoramento da segmentação semântica. O primeiro arcabouço usa super-resolução como um pré-processamento para a tarefa de segmentação semântica. O segundo arcabouço é uma abordagem unificada que treina as duas redes ao mesmo tempo enquanto compartilha suas funções de erro. Foram executados um conjunto extensivo de experimentos em dados de sensoriamento remoto com natureza e propriedades distintas. Para o conjunto de dados agriculturais de mapeamento de café, que contém apenas duas classes (café e não-café), o uso de imagens de baixa resolução alcançou apenas 50% de acurácia normalizada com taxa de aumento de 8 vezes. O arcabouço em dois estágios na mesma condição aumentou esse valor para 72%. O arcabouço unificado aumentou ainda mais esse valor para 77%, comparado aos 81% com dados de alta resolução. Para o conjunto de dados urbano de Vaihingen, usar super-resolução no arcabouço de dois estágios aumentou a acurácia de segmentação de carros de 19% para 58% com taxa de aumento de 8 vezes, enquanto o arcabouço unificado alcançou 65%. Nesse caso, com dados de alta resolução, a acurácia foi de 69%, o que não está distante dos resultados de super-resolução. Ambos os casos são exemplos de como super-resolução é capaz de recuperar detalhes de textura importantes (para plantações de café, por exemplo) e também é capaz de fazer ficarem mais claros objetos que eram difíceis de enxergar em uma representação de baixa resolução (como os carros). Os resultados mostram que super-resolução é efetiva para melhorar o desempenho de segmentação semântica em imagens aéreas de baixa resolução. Super-resolução não apenas é melhor que interpolação não-supervisionada, como também alcança resultados de segmentação semântica comparáveis a dados de alta resolução. Mesmo com pouco dado de treinamento, o uso dos arcabouços alcançou resultados melhores que usando interpolação bicúbica. Dessa forma, o uso de super-resolução se provou ser mais efetivo do que aplicar imagens de baixa resolução em uma rede neural de segmentação semântica. Isso é verdade especialmente para altos fatores de degradação, os quais são os casos em que super-resolução supera mais o desempenho de se usar diretamente dados de baixa resolução.CNPq - Conselho Nacional de Desenvolvimento Científico e TecnológicoengUniversidade Federal de Minas GeraisPrograma de Pós-Graduação em Ciência da ComputaçãoUFMGBrasilICX - DEPARTAMENTO DE CIÊNCIA DA COMPUTAÇÃOComputação – TesesVisão por computador – TesesSuper-resolução – TesesSensoriamento remoto – TesesSegmentação semântica – TesesRemote sensingSuper resolutionSemantic segmentationMapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution imagesMapeando o invisível: explorando super-resolução para segmentação semântica em imagens de baixa resoluçãoinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisinfo:eu-repo/semantics/openAccessreponame:Repositório Institucional da UFMGinstname:Universidade Federal de Minas Gerais (UFMG)instacron:UFMGORIGINALdissertacao_final_review2.pdfdissertacao_final_review2.pdfapplication/pdf19505437https://repositorio.ufmg.br/bitstream/1843/36706/3/dissertacao_final_review2.pdfbac7e982f6e0a6c2f4c0bc4d60a1a6b1MD53LICENSElicense.txtlicense.txttext/plain; charset=utf-82118https://repositorio.ufmg.br/bitstream/1843/36706/4/license.txtcda590c95a0b51b4d15f60c9642ca272MD541843/367062021-07-09 12:03:16.943oai:repositorio.ufmg.br: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ório de PublicaçõesPUBhttps://repositorio.ufmg.br/oaiopendoar:2021-07-09T15:03:16Repositório Institucional da UFMG - Universidade Federal de Minas Gerais (UFMG)false |
dc.title.pt_BR.fl_str_mv |
Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images |
dc.title.alternative.pt_BR.fl_str_mv |
Mapeando o invisível: explorando super-resolução para segmentação semântica em imagens de baixa resolução |
title |
Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images |
spellingShingle |
Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images Matheus Barros Pereira Remote sensing Super resolution Semantic segmentation Computação – Teses Visão por computador – Teses Super-resolução – Teses Sensoriamento remoto – Teses Segmentação semântica – Teses |
title_short |
Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images |
title_full |
Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images |
title_fullStr |
Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images |
title_full_unstemmed |
Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images |
title_sort |
Mapping the unseen: exploiting super-resolution for semantic segmentation in low-resolution images |
author |
Matheus Barros Pereira |
author_facet |
Matheus Barros Pereira |
author_role |
author |
dc.contributor.advisor1.fl_str_mv |
Jefersson Alex dos Santos |
dc.contributor.advisor1Lattes.fl_str_mv |
http://lattes.cnpq.br/2171782600728348 |
dc.contributor.referee1.fl_str_mv |
George Luiz Medeiros Teodoro |
dc.contributor.referee2.fl_str_mv |
André Vital Saúde |
dc.contributor.referee3.fl_str_mv |
Wesley Nunes Gonçalves |
dc.contributor.authorLattes.fl_str_mv |
http://lattes.cnpq.br/5014941996905154 |
dc.contributor.author.fl_str_mv |
Matheus Barros Pereira |
contributor_str_mv |
Jefersson Alex dos Santos George Luiz Medeiros Teodoro André Vital Saúde Wesley Nunes Gonçalves |
dc.subject.por.fl_str_mv |
Remote sensing Super resolution Semantic segmentation |
topic |
Remote sensing Super resolution Semantic segmentation Computação – Teses Visão por computador – Teses Super-resolução – Teses Sensoriamento remoto – Teses Segmentação semântica – Teses |
dc.subject.other.pt_BR.fl_str_mv |
Computação – Teses Visão por computador – Teses Super-resolução – Teses Sensoriamento remoto – Teses Segmentação semântica – Teses |
description |
High-resolution aerial images are desirable for most of the deep-based remote sensing applications. This type of data, however, is not always accessible or affordable. On the other hand, coarse resolution remote sensing images, such as LANDSAT and MODIS, are easily found in public open repositories and, therefore, are widely used in many studies. The problem is that the amount of spacial information compressed into one single pixel in a low-resolution representation can compromise pattern recognition algorithms. Thus, the use of coarse-resolution data for automatic creation of thematic maps is very restricted since most of the deep-based semantic segmentation (a.k.a dense labeling) approaches are only suitable for subdecimeter data. Super-resolution is a classic computer vision problem that aims to restore the quality of degraded, low-resolution images. In this work, we design two frameworks in order to evaluate the effectiveness of deep-based super-resolution in the semantic segmentation of low-resolution remote sensing images. Our objective is to evaluate how effective is deep-based super-resolution to different levels of degradation, how it compares to unsupervised bicubic interpolation and if it is able to reconstruct small objects and, consequently, contribute to semantic segmentation improvement. The first framework uses super-resolution as a pre-processing step for the semantic segmentation task (two-stage framework). The second framework is an end-to-end approach that trains both networks at the same time while sharing their losses. We carried out an extensive set of experiments on remote sensing datasets with distinct nature and properties. For the agricultural dataset of coffee mapping, which only contains two labels (coffee and non-coffee), the use of low-resolution images achieved only 50% normalized accuracy with 8x up-scaling factor. The two stage framework with super-resolution in the same condition increased this value to 72%. The end-to-end framework further increased the value to 77%, compared to 81% of high-resolution data. For the urban dataset of Vaihingen, using super-resolution in the two stage framework increased the accuracy of car segmentation from 19% to 58% with 8x up-scaling factor, while the end-to-end framework achieved 65%. In this case, with high-resolution data, the accuracy was 69%, which is not far from the super-resolution result. Both cases are examples of how super-resolution is able to recover important texture details (for coffee crops, for example) and is also able to make more discernible small objects that were difficult to see in a low-resolution representation (such as cars). The results show that super-resolution is effective to improve semantic segmentation performance on low-resolution aerial imagery. It not only outperforms unsupervised interpolation but also achieves semantic segmentation results comparable to high-resolution data. Even with a few training data, the use of the frameworks still achieved better results than bicubic interpolation. Thus, using super-resolution has proven to be a more effective approach than directly inputting low-resolution images to a semantic segmentation network. This is especially true for high degrading factors, which are the cases that super-resolution surpasses more the performance of low-resolution data. |
publishDate |
2019 |
dc.date.issued.fl_str_mv |
2019-11-11 |
dc.date.accessioned.fl_str_mv |
2021-07-09T15:03:16Z |
dc.date.available.fl_str_mv |
2021-07-09T15:03:16Z |
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/1843/36706 |
url |
http://hdl.handle.net/1843/36706 |
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.publisher.none.fl_str_mv |
Universidade Federal de Minas Gerais |
dc.publisher.program.fl_str_mv |
Programa de Pós-Graduação em Ciência da Computação |
dc.publisher.initials.fl_str_mv |
UFMG |
dc.publisher.country.fl_str_mv |
Brasil |
dc.publisher.department.fl_str_mv |
ICX - DEPARTAMENTO DE CIÊNCIA DA COMPUTAÇÃO |
publisher.none.fl_str_mv |
Universidade Federal de Minas Gerais |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da UFMG instname:Universidade Federal de Minas Gerais (UFMG) instacron:UFMG |
instname_str |
Universidade Federal de Minas Gerais (UFMG) |
instacron_str |
UFMG |
institution |
UFMG |
reponame_str |
Repositório Institucional da UFMG |
collection |
Repositório Institucional da UFMG |
bitstream.url.fl_str_mv |
https://repositorio.ufmg.br/bitstream/1843/36706/3/dissertacao_final_review2.pdf https://repositorio.ufmg.br/bitstream/1843/36706/4/license.txt |
bitstream.checksum.fl_str_mv |
bac7e982f6e0a6c2f4c0bc4d60a1a6b1 cda590c95a0b51b4d15f60c9642ca272 |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da UFMG - Universidade Federal de Minas Gerais (UFMG) |
repository.mail.fl_str_mv |
|
_version_ |
1797971093795897344 |