Productive Crop Field Detection

• Ano de defesa: 2024
• Autor(a) principal: Garcia do Nascimento, Eduardo
• Orientador(a): Almeida, Tiago
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus Sorocaba
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação - PPGCC-So
• Palavras-chave em Português: Agricultura de Precisão; Detecção de talhões; Aprendizado contrastivo
• Palavras-chave em Inglês: Precision Agriculture; Crop field detection; Contrastive learning
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::SISTEMAS DE COMPUTACAO; CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; CIENCIAS EXATAS E DA TERRA::GEOCIENCIAS::GEOFISICA::SENSORIAMENTO REMOTO
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/19206

In precision agriculture, detecting productive crop fields is an essential practice that allows the farmer to evaluate operating performance separately and compare different seed varieties, pesticides, and fertilizers. However, manually identifying productive fields is often time-consuming, costly, and subjective. Previous studies explore different methods to detect crop fields using advanced machine learning algorithms to support the specialists’ decisions, but they often lack good quality labeled data. In this context, we propose a framework for productive crop field detection based on high-quality dataset generated by machine operation combined with Sentinel-2 images tracked over time. As far as we know, it is the first one to overcome the lack of labeled samples by using this combination of techniques. In sequence, we present three methods, based on state-of-the-art supervised and self-supervised methods, selected according to the dataset characteristics, to detect productive crop fields. Finally, we demonstrate high accuracy results in Positive Unlabeled learning, which perfectly fits the problem where we have high confidence in the positive samples. Finally, best performances have been found with Contrastive Learning, given its ability to augment data, allowing the model to be trained with a larger dataset considering the artificially created samples.