High-Accuracy Plant Leaf Disease Classification Using Transfer Learning and Data Augmentation

This research aims to enhance plant leaf disease classification accuracy using advanced machine learning techniques, focusing on grape leaves with diseases like Black measles and Isariopsis leaf spot. We trained and evaluated two key CNN models: our custom CNN and InceptionResNetV2, using the ‘Plant Village Dataset.‘ Preprocessing included standardization and data augmentation. The augmented InceptionResNetV2 model achieved remarkable results: 98.4% training accuracy, 99.5% validation accuracy, and an impressive 99.7% testing accuracy. Comparative analysis with the ‘UnitedModel’ from the base paper highlighted our model's superior testing accuracy. These findings have significant practical implications, offering the potential for precise disease detection, thereby facilitating suitable interventions and sustainable agricultural practices. This research holds practical significance for accurate disease detection, enabling timely interventions in agriculture. It advances plant leaf disease classification, offering a highly accurate model with robust generalization capabilities.