Aplicação de machine learning e sensoriamento remoto para identificação e monitoramento de lavouras de café arábica infestadas por nematoides das galhas

Abstract

Coffee farming plays a significant role in Brazil's economy and culture, especially in Minas Gerais, the largest coffee-producing state. However, production faces serious challenges, such as infestations by root-knot nematodes (Meloidogyne spp., RKN) which significantly impact productivity. This study focuses on the use of machine learning technologies, with an emphasis on the Maximum Entropy (MaxEnt) model and remote sensing, to identify and monitor areas affected by RKN. Spectral data from Landsat-8 satellite images were utilized, along with advanced field and laboratory validation methods, where Meloidogyne species were identified using molecular markers. The results demonstrated the effectiveness of MaxEnt in generating potential distribution maps, enabling more efficient integrated management, as Meloidogyne species were identified in most samples. Furthermore, this research contributed to advancing knowledge on the use of technological tools in agriculture, with significant implications for sustainability and strategic planning in Brazilian coffee farming.