Enhancing Plant Leaf Disease Classification Using Transfer Learning and Ensemble Techniques

This research paper presents a novel approach to plant leaf disease classification using transfer learning and ensemble techniques. The primary objective is to recover the accuracy and efficiency of disease identification to support effective agricultural disease management. The methodology involves the utilization of three pre-trained “Convolutional Neural Network” (CNN) models—VGG16, ResNet50, and VGG19—as feature extractors for disease classification. Data preprocessing techniques, including data augmentation, standardize the dataset and enhance generalization capabilities. Subsequently, using ensemble learning, we combine the predictions of the various models to form the hybrid model. Important results show that the suggested method works. The standalone CNN models achieve high training and testing accuracies, showcasing their capability to learn intricate disease-specific features. The hybrid model, resulting from ensemble learning, outperforms the standalone models, surpassing 99% accuracy in both training and testing phases. These results highlight the potential of transfer learning and ensemble techniques in enhancing disease identification accuracy.