@InProceedings{CollegioGuimDalP:2023:DeRoRu,
               author = "Collegio, Gustavo Rota and Guimar{\~a}es Filho, Antonio Gaudencio 
                         and Dal Poz, Aluir Porfirio",
          affiliation = "{Universidade Estadual Paulista (UNESP)} and {Universidade 
                         Estadual Paulista (UNESP)} and {Universidade Estadual Paulista 
                         (UNESP)}",
                title = "Detec{\c{c}}{\~a}o de rodovias rurais em imagens orbitais 
                         atrav{\'e}s do emprego de redes neurais convolucionais",
            booktitle = "Anais...",
                 year = "2023",
               editor = "Gherardi, Douglas Francisco Marcolino and Arag{\~a}o, Luiz 
                         Eduardo Oliveira e Cruz de and Sanches, Ieda DelArco",
                pages = "e155847",
         organization = "Simp{\'o}sio Brasileiro de Sensoriamento Remoto, 20. (SBSR)",
            publisher = "Instituto Nacional de Pesquisas Espaciais (INPE)",
              address = "S{\~a}o Jos{\'e} dos Campos",
             keywords = "Detec{\c{c}}{\~a}o de rodovias, RNC/U-Net, 
                         segmenta{\c{c}}{\~a}o sem{\^a}ntica,detection, CNN/U-Net, 
                         semantic segmentation.",
             abstract = "A detec{\c{c}}{\~a}o de rodovias por meio de imagens orbitais 
                         apresenta relev{\^a}ncia significativa na comunidade 
                         cient{\'{\i}}fica em fun{\c{c}}{\~a}o das diversas 
                         aplica{\c{c}}{\~o}es que as concerne, tais como: planejamento 
                         urbano, atualiza{\c{c}}{\~a}o de banco de dados 
                         cartogr{\'a}ficos etc. O m{\'e}todo proposto se baseia em uma 
                         Rede Neural Convolucional (RNC), daqui em diante identificada como 
                         RNC/U-Net, que visa a detec{\c{c}}{\~a}o de rodovias em 
                         regi{\~o}es rurais, por meio de um processo denominado 
                         segmenta{\c{c}}{\~a}o sem{\^a}ntica. A {\'a}rea teste usada 
                         para avaliar o m{\'e}todo se localiza no estado do Mato Grosso. A 
                         RNC/U-Net alcan{\c{c}}ou 58,44% de recall e 49,65% de precision, 
                         com 36,26% de Intersection-Over-Union. Os resultados obtidos 
                         mostraram que a arquitetura {\'e} eficiente na 
                         detec{\c{c}}{\~a}o de rodovias rurais; no entanto para aquelas 
                         de car{\'a}ter radiom{\'e}trico e geom{\'e}trico similar com 
                         outros alvos, a RNC/U-Net ainda {\'e} pass{\'{\i}}vel de 
                         aperfei{\c{c}}oamentos e adapta{\c{c}}{\~o}es, visando 
                         contribui{\c{c}}{\~a}o direta na segmenta{\c{c}}{\~a}o das 
                         rodovias. ABSTRACT: Road detection through orbital images is 
                         extremely relevant in the scientific community due to the various 
                         applications that concern them, such as urban planning, 
                         cartographic databases updating etc. The proposed method is based 
                         on a Convolutional Neural Network (CNN), from here on identified 
                         as CNN/U-Net, that aims at detecting roads is rural regions, 
                         through a process that is known as semantic segmentation. The test 
                         area used to evaluate the proposed method is localized in Mato 
                         Grosso state. The RNC/U-Net reached 58.44% of recall and 49.65% of 
                         precision, with 36.26% of Intersection-Over-Union. The results 
                         obtained showed that the architecture was efficient in detecting 
                         rural roads, however, for those of radiometric and geometric 
                         character similar to other targets, the RNC/U-Net is still subject 
                         to improvements and adaptations, aiming at a direct contribution 
                         to the segmentation of roads.",
  conference-location = "Florian{\'o}polis",
      conference-year = "02-05 abril 2023",
                 isbn = "978-65-89159-04-9",
             language = "pt",
         organisation = "Instituto Nacional de Pesquisas Espaciais (INPE)",
                  ibi = "8JMKD3MGP6W34M/4939P82",
                  url = "http://urlib.net/ibi/8JMKD3MGP6W34M/4939P82",
           targetfile = "155847.pdf",
                 type = "Cartografia e fotogrametria",
        urlaccessdate = "12 maio 2024"
}