YOLO Based Deep Learning Framework for Cotton Leaf Disease Detection in Smart Farming Systems

Cotton is the backbone of the global textile economy, yet it is highly vulnerable to diseases that cause substantial yield and quality reduction. Conventional manual detection is time-consuming, prone to errors, and cannot offer real-time data for monitoring agriculture at a large scale. In this paper, we present a deep learning method based on YOLOv10 for detecting and classifying the most common diseases in cotton leaves. The proposed scheme is designed with a dataset of 1,710 high-quality images captured in various natural conditions. To improve the robustness and generalization of the model to different backgrounds and illumination conditions, more data augmentation techniques were applied. By using YOLOv10’s enhanced backbone and the scale-aware neck structure, this framework is able to efficiently process computations while maintaining a very high degree of precision in diagnosis. Experimental results show that the proposed YOLOv10 model has a maximum detection accuracy of 90%, and greatly including other versions such as YOLOv5, YOLOv7, and YOLOv8. In addition, the model displayed superior performance compared with classical classifiers (SVM and KNN). The proposed framework is well-suited for real-time applications on mobile platforms and unmanned aerial vehicles (UAVs) in smart farming systems. This system allows for precision pesticide use a disease management which contribute to sustainable high-yield cotton production. Which outperforms normal models. The results demonstrate the feasibility that the suggested method can reform the technique of crop disease detection for the intervention in time to keep yield up.