LeafData.zip

Plant diseases pose a significant threat to agricultural sustainability by reducing crop productivity. The rapid and accurate diagnosis and management of these diseases are crucial. Recent advancements in artificial intelligence have facilitated the development of automated systems for disease detection. This study aims to enhance the classification and severity estimation of diseases on coffee leaf images. The proposed method utilizes Enhanced Multivariance Product Representation (EMPR) to decompose the image into its components prior to classification. A new image is generated from selected components, and its contrast is enhanced using High-Dimensional Model Representation (HDMR) to emphasize the diseased areas of the leaves. The performance of popular convolutional neural network (CNN) architectures, including AlexNet, VGG16, and ResNet50, is tested. The results indicate that VGG16 achieves the highest classification accuracy, approximately 96%, while all models exhibit strong performance in estimating disease severity levels with accuracies exceeding 85%. Notably, the ResNet50 model attains accuracy levels surpassing 90%.