Hybrid Deep Convolutional Network for Accurate Grapevine Leaf Disease Classification

Grapevine leaf diseases significantly affect vineyard productivity and fruit quality, making rapid and accurate disease identification essential for precision agriculture and sustainable crop management. This study presents an intelligent deep learning framework for automated grapevine disease classification using a customized Convolutional Neural Network (CNN). The proposed model categorizes grapevine leaf images into four major classes: Black Rot, ESCA, Healthy, and Leaf Blight. A comprehensive image dataset was prepared and divided into training, validation, and testing subsets with ratios of 70%, 15%, and 15%, respectively, to ensure balanced learning and unbiased evaluation. Before training, all images were resized to 224 × 224 pixels and normalized to enhance feature consistency. The CNN architecture incorporates multiple convolutional and max-pooling layers for effective hierarchical feature extraction, followed by dense layers with dropout regularization to reduce overfitting and improve generalization capability. Experimental evaluation demonstrates strong classification performance, achieving approximately 99% training accuracy, 96% validation accuracy, and 95% testing accuracy. Furthermore, confusion matrix analysis and class-wise performance metrics confirm the robustness and reliability of the proposed approach across all disease categories. The developed system offers an efficient, scalable, and accurate solution for early grapevine disease detection, supporting smart viticulture and automated agricultural monitoring applications.