YOLO-based Deep Learning Framework for Cotton Leaf Deficiency Diagnosis

Cotton, a globally vital fiber crop, is vulnerable to numerous nutrient deficiencies and diseases that drastically affect yield and fiber quality. Traditional monitoring techniques rely heavily on manual scouting, which is often inefficient, error-prone, and lacks scalability. Unmanned Aerial Vehicles (UAVs), integrated with deep learning models, offer a promising solution for real-time, high-resolution crop diagnostics. This study evaluates four YOLO-based object detection models—YOLOv8n, YOLOv8s, YOLOv10n, and YOLOv10s—on UAV-acquired imagery for identifying nutrient-deficient cotton leaves. YOLOv10s achieved the best performance (F1-score: 0.83, [email protected]: 0.885), highlighting its potential for deployment in precision agriculture systems. We further contextualize our findings within recent advances in deep learning, UAV design, and agricultural image analytics. We also explore the feasibility of applying few-shot learning to improve detection in low-data regimes, enabling broader deployment across crop types.