Deep Learning for Accurate Vineyard Pathology Detection

This research paper introduces an innovative method for detecting diseases in the grapevine leaves using Faster R-CNN, a cutting-edge deep learning technology is vital for timely disease management in vineyards. The study focuses on identifying multisymptomatic soil-borne diseases like Grape Esca, Black rot, and Isariopsis, which may result in severe consequences such as leaf loss or plant disappearance. Unlike conventional methods such as Generative Adversarial Networks (GANs) [1], which are time-consuming and costly, the proposed Vineyard Pathology Detection Network (VPDN) on VGG16 [4] leverages Faster R-CNN’s [2], [3] precision in localizing and classifying infected grape leaves within images. This approach considerably outperforms previous methods, achieving an outstanding 98.75% accuracy rate compared to the model Leaf GAN [1] with 98.70%, demonstrating its effectiveness in accurate and rapid disease detection.