Performance Analysis of YOLOv8, YOLOv9, and YOLOv11 for Corn Leaf Disease Detection

Abstract

Corn is a crucial crop for agricultural yield and food security, yet it faces significant threats from various foliar diseases that can diminish both growth and quality. Conventional visual assessment techniques often require substantial labor and are susceptible to errors in diagnosis. This study introduces a Corn Leaf Disease Detection System that employs Convolutional Neural Networks (CNNs) and evaluates the performance of three models: YOLOv8, YOLOv9, and YOLOv11. The methodology involves capturing high-resolution images of corn leaves afflicted by diseases such as Northern Leaf Blight and Common Rust. These images undergo a pre-processing phase to enhance clarity and are standardized for input into the detection framework. CNN is employed for intricate classification tasks, while YOLOv11 is utilised for real-time disease detection. The dataset comprises 3,000 images, which are augmented to expand the training set to 6,000 samples. Among the evaluated models, YOLOv11 demonstrated superior performance, achieving an F1-score of 0.93, with precision at 0.94 and recall at 0.92 by epoch 100. These findings highlight the system's operational efficiency and robustness in effectively detecting corn leaf diseases.