Abstract
This paper introduces a novel AI-based vision system aimed at enhancing the precision and automation of microgreen basil cultivation in a vertical farming environment. By integrating a high-resolution camera within a controlled grow box, real-time images are captured and analyzed using the Ultralytics YOLO11 model to detect microgreens and measure their height in non-invasive ways. A reference coin with a known diameter is employed to convert pixel dimensions into real-world units, thereby ensuring accurate and consistent height measurements. Model performance is rigorously evaluated against manual measurements across varying environmental conditions—such as lighting, angles, and backgrounds—to demonstrate robustness and reliability. Quantitative assessments reveal high F1 scores, precision, and recall values for both the microgreen and reference object classes, underscoring the model’s strong detection capability. These findings highlight the potential for substantial labor savings, improved resource utilization, and more informed decision-making in agricultural management. By offering an efficient, scalable framework for real-time plant monitoring, this research paves the way for broader integration into IoT-driven agricultural systems. Ultimately, the proposed approach contributes significantly to advancing sustainable and datadriven practices in precision agriculture and plant science research.