Performance Analysis of YOLO Models for Beetroot Crop Disease Detection

In agriculture, technological advancements are essential to meeting the growing global food demand, particularly in the early detection and control of crop diseases. This work presents an automated disease detection system for Cercospora leaf spot in beetroot crops, utilizing the YOLO object detection model. Deep learning models YOLOv5s and YOLOv8x are used for processing. Cercospora beticola, a fungal pathogen, significantly reduces beetroot yield and quality, with its impact worsened by environmental conditions. A detailed comparative analysis of YOLOv5s and YOLOv8x is conducted, demonstrating YOLOv8x’s superior performance for cercospora disease detection in beetroots. The experimental results indicate that YOLOv8x outperforms YOLOv5s, achieving a precision of 0.924, recall of 0.821, F1-score of 0.869, and mAP50 of 0.944, compared to YOLOv5s, which attained a precision of 0.814, recall of 0.857, F1-score of 0.834, and mAP50 of 0.884. The improved detection accuracy of YOLOv8x demonstrates its effectiveness in disease detection. By automating disease detection using a deep learning approach, this system enhances agricultural productivity, minimizes crop losses, and promotes sustainable beetroot farming through precise and timely intervention.