Detalhes bibliográficos
Ano de defesa: |
2024 |
Autor(a) principal: |
Vitória Fátima Bernardo |
Orientador(a): |
Gustavo de Faria Theodoro |
Banca de defesa: |
Não Informado pela instituição |
Tipo de documento: |
Dissertação
|
Tipo de acesso: |
Acesso aberto |
Idioma: |
por |
Instituição de defesa: |
Fundação Universidade Federal de Mato Grosso do Sul
|
Programa de Pós-Graduação: |
Não Informado pela instituição
|
Departamento: |
Não Informado pela instituição
|
País: |
Brasil
|
Palavras-chave em Português: |
|
Link de acesso: |
https://repositorio.ufms.br/handle/123456789/11114
|
Resumo: |
To prevent yield losses caused by pathogens, monitoring techniques using hyperspectral images have shown promise, given their capacity to accurately and early diagnose diseases. In this context, the objectives of this study were: to identify differences in the spectral signature of different severities of target spot in cotton; to identify the most accurate machine learning algorithm (ML) for classifying levels; and to determine which sample size (40, 60, 80, or 100) ensured the best accuracy. The experiment was conducted during the 2023/24 agricultural year, in Costa Rica, MS. To obtain the severity levels of target spot (N1 – healthy leaves; N2 – leaves with 1 to 9% severity; N3 – leaves with 19 to 37%; N4 – severity of 53% or higher), experimental field plots were set up, where four different fungicide combinations were applied. At the F14 stage, 100 leaf samples were collected at each severity level in order to perform hyperspectral readings using the FieldSpec spectroradiometer. These data were subjected to ML analysis using six different algorithms. In addition, the wavelengths obtained were separated into 28 bands and then submitted to principal component analysis. It was found that the spectral curve exhibited distinct signatures for the disease severity levels. Among the models analyzed, SVM showed the highest accuracy in classification. Additionally, the sample sizes of 80 and 100 leaves showed greater accuracy. |