Inteligência artificial na análise da maturação do café e a relação com a produtividade bienal da cultura

Abstract

Coffee plays a crucial role in Brazil’s economy, where the adoption of new technologies enables strategies aimed at improving production efficiency. This research aimed to highlight the importance of the harvest timing on the productivity of processed coffee and its possible relationship with final fruit quality during the 2024 and 2025 seasons, as well as the use of artificial intelligence (Yolo_m8) for fruit classification and process automation. The study was conducted in a 27-ha irrigated coffee area located at 980 m of altitude, under a drip irrigation system. A grid of 50 sampling points was delimited using QGIS software for data collection. At each point, four plants were selected for manual harvest, and photographs of the harvested material were taken. The images were obtained with a 50-megapixel mobile phone camera at 1.70 m above ground. For productivity determination, the grains were dried to approximately 11.5% moisture and processed. Image analysis allowed for the estimation of the percentages of fruits at different maturation stages at harvest time. Spatial dependency maps were generated for all collected attributes through geostatistical analysis and QGIS. The results indicated spatial dependence for both processed coffee yield and fruit maturation stages. The relationship observed between these attributes at harvest time may influence productivity and quality, especially when a higher proportion of green fruits was present.