GAN and MobileNet Based Classification of Disease Severity of Septoria Leaf Spot in Tomato Crop

The innovative framework for the precise classification of Septoria leaf spot disease severity in tomato crops, integrating MobileNet and Generative Adversarial Networks (GANs) to enhance the accuracy and accessibility of disease severity detection. MobileNet, with its optimized architecture for mobile devices, serves as the foundation of the system, ensuring that farmers can access real-time disease severity assessments directly in the field. GANs are employed to generate synthetic images that enrich the training dataset, thereby enhancing the model's robustness and its ability to generalize to unseen data without the need for extensive annotated datasets. The project utilizes the PlantVillage dataset, focusing on disease severity of Septoria leaf spot in tomato plant, and evaluates the performance of CNN, MobileNet, and MobileNetV2 architectures, both with and without GAN-augmented data. The results indicate that the MobileNetV2 model, when combined with GAN-generated data, significantly outperforms other models in terms of test accuracy, achieving a remarkable 91.01% accuracy while maintaining computational efficiency. This approach underscores the potential of deep learning technologies to revolutionize plant disease management, empowering farmers with actionable insights to mitigate crop losses and contribute to global food security.