Grape Leaf Disease Detection And Classification Using Machine Learning

Plant diseases present a major threat to global food security, affecting agricultural productivity and livelihoods across the globe. We implemented two different model architectures for comparison: a custom CNN model designed specifically for plant disease detection and the VGG16 pretrained model,. To optimize the performance of these models, we tested three popular optimization techniques: RMSprop, Adam, and Nadam, SGD. The performance of the custom CNN model and the VGG16 model when trained with different optimization algorithms. Across all experiments, the Adam optimizer consistently achieved the highest accuracy, with an average accuracy of 96%. The Nadam optimizer, while also exhibiting strong performance, yielded slightly lower accuracy, averaging $\mathbf{9 1 \%}$. In contrast, the SGD optimizer lagged behind significantly, achieving an average accuracy of only 55%. These results suggest that the choice of optimization algorithm has a significant impact on the convergence speed and final accuracy of machine learning models for plant leaf disease detection.