Grape Leaf Disease Prediction and Management System

I. ABSTRACT The grapevine industry faces numerous challenges, including the prevalence of diseases that can severely impact crop yield and quality. Timely identification of grapevine diseases is critical for effective management and prevention. Traditional methods of disease detection, relying on visual inspection by experts, are often time-consuming, inconsistent, and prone to human error. To address this challenge, we propose a state- of-the-art Grape Disease Detection System using YOLOv8 (You Only Look Once), an advanced deep learning-based object detection algorithm, for real-time identification and classification of grapevine diseases. The system leverages computer vision techniques to detect various grapevine diseases from images of grape leaves. It uses a dataset of labelled grapevine leaf images, collected under various grapevine leaves affected by diseases such as ESCA, Leaf Blight, and other common fungal infections. The dataset undergoes preprocessing steps such as resizing, normalization, and augmentation to ensure robustness and generalization of the model. The model's architecture enables it to detect and classify diseases in images captured by cameras or smartphones with high accuracy and speed. Once trained, YOLOv8 processes new images, detects diseased regions, and classifies them into disease categories based on learned features. The system provides visual feedback by drawing bounding boxes around diseased areas and labels them with the disease type and the model's confidence score. It also generates real-time alerts and recommendations for treatment based on the detected disease. The system's workflow begins with the acquisition of images, followed by preprocessing, disease detection, classification, and results display. It can be deployed as a standalone mobile application or integrated with existing vineyard monitoring systems.