Deep Learning-based VGG16, VGG19, and ResNet Models for Grapevine Disease Classification

Identifying grape plant disease is very important to keep the vineyard healthy and make sure we can grow good grapes. This study looks into using deep learning models like VGG16, VGG19 and ResNet to automatically find out problems in grapevine health. The models were trained on a wide set of 9027 images from Kaggle. These pictures showed different types of disease, including powdery mildew and downy mildew among others. We used a learning rate of 0.01 to help the model come together during training faster. The results show that all three models work very well in finding grapevine diseases. VGG19 was the best one, reaching a correct rate of 98.39% at stage number 24. This better performance comes from its more complex design, allowing it to catch complicated disease parts. VGG16 also did very well, reaching 93.46% correctness, while ResNet had a strong accuracy of 89.33%. The results show that using machine learning deep models can help find grapevine diseases easily. This gives farmers a good tool to manage their vineyards well in advance. The side-by-side look shows the special good points of each design. This study helps to mix artificial intelligence with grape farming. It gives tips that could be helpful in choosing how to handle diseases better. Later research may look at easier ways to understand models, how well they work in different areas where vines grow, and if they can be used for more things. This is important for the grape and wine industry.