Detecção de ninhos de formigas do gênero Atta utilizando visão computacional e quadrirrotores

• Ano de defesa: 2023
• Autor(a) principal: Almeida, Tayná Bertacine de
• Orientador(a): Hernandes, André Carmona
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Engenharia Elétrica - PPGEE
• Palavras-chave em Português: Sensoriamento remoto; Agricultura de precisão; Visão computacional; Robótica aérea; Formigueiro
• Palavras-chave em Inglês: Remote Sensing; Precision agriculture; Aerial robotics; Computer vision; Ant nest
• Área do conhecimento CNPq: ENGENHARIAS::ENGENHARIA ELETRICA
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/18830

In the last 40 years, Brazilian agricultural production and productivity have had significant increases. Agribusiness has been strongly recognised in the growth of the Brazilian Gross Domestic Product, which has resulted in investments and technological advances in the agricultural sector. With these technological advances came precision agriculture. As part of the advances in precision agriculture, Unmanned Aircraft are gaining more and more space, given their flexibility of use and diversity of applications. At the same time, the use of computer vision and machine learning brings analysis technologies that further optimise processes, such as pasture mapping and pest detection. And when it comes to crop pests, leaf-cutter ants come to mind. The direct and indirect damage that leaf-cutting ants can cause amounts to billions of dollars worldwide. The leaf-cutting ant, known as Sauva, is one of the insects that causes most damage to pastures in southeastern Brazil, even competing with cattle for grass. Pest control is of paramount importance for the growth of agricultural production, therefore, this article presents the proposal for the development of an algorithm capable of detecting anthills. Based on recent tech nological advances and cost reduction, the proposed detection was performed using an unmanned aerial vehicle with an integrated conventional RGB camera and an algorithm, based on computer vision and machine learning. The proposed classification algorithm was Random Forest and it presented an accuracy of approximately 70%. The applica tion of computer vision and quadrotors in the detection of ant nests represented a good opportunity to boost precision agriculture and pest control in the field.