AgriEngineering

Accurate seed singulation is critical for uniform crop establishment and yield optimization in precision agriculture. This study presents the development and evaluation of a Hall sensor-based Seed Miss Prevention System (SMPS) integrated into a long-belt finger-clip precision metering device for corn (‘Dekalb DKC5032’) and sunflower (‘Astana’). The system utilizes neodymium magnets mounted on seed-picking fingers to trigger a Hall sensor that detects missed seeds in real time and initiates immediate compensation. Laboratory tests across rotational speeds from 10 to 80 rpm showed that the SMPS significantly reduced seed misses, especially within the 10–30 rpm range, where near-perfect singulation was achieved for corn (Miss Index < 0.01). For sunflower, although performance at very low speeds was limited by mechanical variability, the SMPS effectively reduced the miss index by up to 50 % at medium speeds. Statistical analysis (Tukey HSD) confirmed significant improvements in single and double miss prevention across both crops. The proposed Hall sensor-based approach demonstrated a robust, cost-effective, and dust-resistant solution for enhancing seed placement accuracy, providing a strong foundation for the development of intelligent and adaptive precision seeding systems.

The coffee leaf miner (Leucoptera coffeella) is a major pest of coffee crops and can cause significant economic losses. Early monitoring is essential to support decision-making for its control. This study aimed to evaluate the potential of fuzzy logic for detecting leaf miner infestation using a 2.5-year historical series of Sentinel-2A satellite images processed on the Google Earth Engine platform. Field monitoring of coffee leaf miner infestation was carried out at the EPAMIG Experimental Field, located in São Sebastião do Paraíso, Minas Gerais, Brazil. The period evaluated was from September 2022 to April 2025. Vegetation indices were calculated using the Google Earth Engine platform, and a database was built with eight indices (NDVI, EVI, GNDVI, SR, IPVI, NDMI, MCARI, and CLMI) along with coffee leaf miner infestation data. Principal Component Analysis (PCA) was applied to reduce data dimensionality and identify the most relevant indices for distinguishing infested from healthy plants, explaining 90.9% of the total variance in the first two components (PC1 and PC2). The indices CLMI, IPVI, GNDVI, and MCARI showed the greatest contribution to class separation. A fuzzy inference model was implemented based on the mean index values and validated through performance metrics. The results indicated an overall accuracy of 79.1%, a sensitivity (recall) of 86.6%, a specificity of 66.6%, an F1-score of 0.838, a Kappa coefficient of 0.545, and an area under the curve (AUC) of 0.766. These findings confirm the potential of integrating orbital spectral data via Google Earth Engine with fuzzy logic analysis as an efficient tool, contributing to the adoption of more sustainable monitoring practices in coffee farming. The fuzzy logic system received as input the spectral values derived from Sentinel-2A imagery, specifically the indices identified as most relevant by the PCA (CLMI, IPVI, GNDVI, and MCARI). These indices were computed and integrated into the inference model through processing routines developed in the Google Earth Engine platform, enabling a direct connection between satellite-derived spectral patterns and the detection of coffee leaf miner infestation.

Monitoring groundcover diversity in vineyards is a complex task, often limited by the time and expertise required for accurate botanical identification. Remote sensing technologies and AI-based tools are still underutilized in this context, particularly for classifying herbaceous vegetation in inter-row areas. In this study, we introduce a novel approach to classify the groundcover into one of nine categories, in order to simplify this task. Using UAV images to train a convolutional neural network through a deep learning methodology, this study evaluates the effectiveness of different backbone structures applied to a UNet network for the classification of pixels into nine classes of groundcover: vine canopy, bare soil, and seven distinct cover crop community types. Our results demonstrate that the UNet model, especially when using an EfficientNetB0 backbone, significantly improves classification performance, achieving 85.4% accuracy, 59.8% mean Intersection over Union (IoU), and a Jaccard index of 73.0%. Although this study demonstrates the potential of integrating remote sensing and deep learning for vineyard biodiversity monitoring, its applicability is limited by the small image coverage, as data were collected from a single vineyard and only one drone flight. Future work will focus on expanding the model’s applicability to a broader range of vineyard systems, soil types, and geographic regions, as well as testing its performance on lower-resolution multispectral imagery to reduce data acquisition costs and time, enabling large-scale and cost-effective monitoring.