ARCHIVES

Achieving robust agricultural yields depends crit- ically on maintaining optimal crop nutrition, especially the balance of essential macronutrients (Nitrogen, Phosphorus, Potassium) and secondary nutrients (Calcium, Sulphur, Mag- nesium). A shortfall in any of these vital elements immediately manifests as distinct visual pathologies, including alterations in leaf pigmentation, texture, and physical structure. Ex- isting diagnostic methods, such as field expert assessment or detailed chemical analysis, are often slow and lack con- sistency, delaying timely intervention. This research intro- duces an automated, economical, and dependable detection framework that leverages classical Image Processing (IP) for feature extraction alongside supervised Machine Learning (ML) for classification. We derive a comprehensive set of visual features—including RGB color statistics, textural measures (Gray-Level Co-occurrence Matrix), and structural boundary features (Canny/Sobel)—to train a robust Random Forest classifier capable of accurately identifying six specific nutrient deficiencies. This system facilitates accurate, early diagnosis, supporting timely corrective action and maximizing overall farm productivity in the context of precision agriculture.