Hyperparameter Optimization of CNN Models for Plant Leaf Disease Classification Using Optuna

Leaf diseases are a major challenge for agricultural productivity, requiring accurate and efficient detection methods. This research presents an effective method for multi-class classification of leaf diseases in apple and mango plants by leveraging MobileNetV2 and XceptionNet architectures. A diverse dataset enhances model generalizability, while the Optuna framework optimizes hyperparameters for improved performance. Among the tested models, Optuna-optimized MobileNetV2 achieves the highest accuracy of 94.05%. The comparative analysis highlights the advantages of combining advanced CNN architectures with hyperparameter optimization for leaf disease detection. This study highlights the importance of optimized models in tackling complex agricultural issues and demonstrates their potential to enhance data-driven decision-making. The results open avenues for future research focused on scalable and real-time smart farming applications.