Wheat Disease Detection Using YOLO and Drone-Captured Images

Mitigating crop loss and maximizing resource usage in agriculture are contingent upon the timely and precise identification of wheat illnesses. In this paper, we offer a YOLOv8-based method that uses drone-captured images to detect wheat diseases in real time. Four distinctive image classes represent our dataset: Ground, Yellow Rust, Brown Rust, and Healthy. Smart spray drones can target disease management by recognizing non-crop regions in the field, a task made possible in large part by the Ground class. This class allows our model to distinguish between areas that need to be treated and those that don’t, improving chemical spraying accuracy and cutting down on waste. Using data gathered from wheat fields, the model was trained and tested, and it performed excellently in differentiating between crops that were diseased and those that weren’t. The model achieved a highest precision of 0.803 and recall of 0.850 across various classes and the overall performance included a mean average precision of 0.605, demonstrating robust performance in field conditions, which qualifies it for use in agricultural monitoring systems that operate in real-time. This effort is a step toward automated, data-driven precision agriculture, which will assist farmers in allocating resources and managing diseases in a timely and effectively manner.