VGG16-based Multiclass Classification of Grapevine Leaf Diseases for Precision Viticulture

Grape is one of the fruit which contains a rich amount of minerals including manganese, potassium, and other vitamins namely C, B, and K, which are essential to prevent osteoporosis. With the increase in world population, the fruit demand has also increased. However, there are various diseases which affect the health of the grapevine leading to loss in its production. In the present study, a pre-trained and fine-tuned VGG16 model has been implemented for precise and accurate grape leaf disease identification. It is crucial to determine the health and growth of grapevines, but the use of advanced computer vision methods can speed up this process. The VGG16 model extracts deep features from the images and allows a faster and non-invasive performance evaluation. In the proposed work, the significance of grape leaf disease detection in viticulture is examined. The study implements the fine-tuned VGG16 model on a Kaggle-based dataset and examines the performance outcomes. The model’s performance results have been identified in terms of accuracy and loss on different numbers of epochs. The results show that with the increase in the epoch count, the accuracy has also increased and reached 98% at epoch 25. The proposed VGG16 model for grape leaf disease recognition is a significant step forward for the viticulture sector, offering a potential approach for automating the evaluation of grape plant health. The proposed work also discusses about the training and testing loss and results that at epoch count 25, the lowest training and testing loss of 0.1466, and 0.1051 have been identified. These findings demonstrate the VGG16 model’s effectiveness in identifying grape leaf diseases by demonstrating its precision and potential to improve both the effectiveness and quality of grapevine management.