@InProceedings{DallaquaSchRosGueRod:2023:ClGêSi,
               author = "Dallaqua, Fernanda Beatriz Jordan Rojas and Schultz, Bruno and 
                         Rosa, Rafael Ant{\^o}nio and Guerra, J{\'u}lio Bandeira and 
                         Rodrigues, Thiago Gon{\c{c}}alves",
          affiliation = "{Visiona Tecnologia Espacial S.A.} and {Visiona Tecnologia 
                         Espacial S.A.} and {Visiona Tecnologia Espacial S.A.} and {Visiona 
                         Tecnologia Espacial S.A.} and {Visiona Tecnologia Espacial S.A.}",
                title = "Classifica{\c{c}}{\~a}o do g{\^e}nero de silvicultura 
                         utilizando s{\'e}ries temporais multi-sensor e aprendizado de 
                         m{\'a}quina",
            booktitle = "Anais...",
                 year = "2023",
               editor = "Gherardi, Douglas Francisco Marcolino and Arag{\~a}o, Luiz 
                         Eduardo Oliveira e Cruz de and Sanches, Ieda DelArco",
                pages = "e155998",
         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 = "Pinus, Eucalipto, LSTM, Classifica{\c{c}}{\~a}o de s{\'e}ries 
                         temporais, Deep learning, Pinus, Eucalyptus, LSTM, Time series 
                         classification, Deep learning.",
             abstract = "No Sul do Brasil tem se tornado comum a introdu{\c{c}}{\~a}o de 
                         esp{\'e}cies de Eucalipto em {\'a}reas que at{\'e} ent{\~a}o 
                         eram plantadas com os g{\^e}neros Pinus spp. e Araucaria sp.. 
                         Atualmente, existe uma necessidade de entendimento do plantio do 
                         Eucalipto no Sul do pa{\'{\i}}s, principalmente para 
                         an{\'a}lises espaciais e de intelig{\^e}ncia de mercado. Este 
                         trabalho teve como objetivo o uso de s{\'e}ries temporais 
                         multi-sensor e t{\'e}cnicas de aprendizado de m{\'a}quina e deep 
                         learning na separa{\c{c}}{\~a}o de Pinus e Eucalipto em duas 
                         fazendas em Tel{\^e}maco Borba PR. Foram utilizadas quatro 
                         t{\'e}cnicas de classifica{\c{c}}{\~a}o: (i) FCN Block, (ii) 
                         t-LSTM, (iii) t-biLSTM e (iv) SVM. A melhor t{\'e}cnica foi a 
                         t-biLSTM, que apresentou um coeficiente Kappa de 0, 88, seguida de 
                         t-LSTM (Kappa = 0, 87), FCN Block e SVM (Kappa = 0, 81). As 
                         t{\'e}cnicas que usaram t-biLSTM e t-LSTM foram iguais entre si e 
                         superiores {\`a}s outras t{\'e}cnicas, ao n{\'{\i}}vel de 95% 
                         de confian{\c{c}}a. ABSTRACT: In southern Brazil, it has become 
                         common to introduce Eucalyptus species in areas that until then 
                         were planted with the genera Pinus spp. and Araucaria sp. 
                         Currently, there is a need to understand the eucalyptus plantation 
                         in the south of the country, mainly for spatial analysis and 
                         market intelligence. This work aimed to use multi-sensor time 
                         series and machine learning and deep learning techniques in the 
                         separation of Pinus and Eucalyptus in two farms in Tel{\^e}maco 
                         Borba - PR. Four classification techniques were used: (i) FCN 
                         Block, (ii) t-LSTM, (iii) t-biLSTM and (iv) SVM. The best 
                         technique was t-biLSTM, which presented a Kappa coefficient of 
                         0.88, followed by t-LSTM (Kappa = 0.87), FCN Block and SVM (Kappa 
                         = 0.81). The techniques that used t-biLSTM and t-LSTM were equal 
                         to each other and superior to the other techniques, at the 95% 
                         confidence level.",
  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/48UQ4LS",
                  url = "http://urlib.net/ibi/8JMKD3MGP6W34M/48UQ4LS",
           targetfile = "155998.pdf",
                 type = "Intelig{\^e}ncia Artificial para Observa{\c{c}}{\~a}o da 
                         Terra",
        urlaccessdate = "11 maio 2024"
}