A Deep Learning Based Fusion Model For Plant Leaf Disease Classification

Agriculture is the backbone of a country's economy. For agricultural sustainability, it is important to accurately and early diagnose if a plant is infected by any disease in order to protect the plant. This paper proposes a deep learning-based fusion model for plant leaf disease classification. Using pre-trained models (ResNet50, VGG16, and EfficientNetB0), the features extracted from images are concatenated and then input into a deep neural network for classification. The proposed work utilizes the publicly available PlantVillage dataset from Kaggle for the leaf disease classification. In this study, 8 balanced classes were used. The proposed fusion model achieves the highest classification test accuracy of 98.25%, with training 99.15%, a loss of 0.0315, and an average 5-fold cross-validation accuracy of 98.7625%. The proposed model is effective and can contribute to agricultural sustainability.