Plant disease detection using a hybrid machine learning model

Abstract

This research presents a hybrid Convolutional Neural Network–Support Vector Machine (CNN-SVM) approach for accurate plant disease detection, integrating CNN’s feature extraction capabilities with SVM’s robust classification performance. The methodology began with data acquisition and preprocessing, including image normalization, augmentation, and resizing to ensure model compatibility and improve generalization. The CNN component was trained to automatically extract discriminative features from plant leaf images, which were subsequently fed into an SVM classifier optimized through hyperparameter tuning. Performance evaluation employed standard metrics, including accuracy, precision, recall, and F1-score, alongside the Receiver Operating Characteristic (ROC) curve analysis. Experimental results demonstrate the hybrid CNN-SVM model’s superiority over standalone CNN and SVM models. The proposed model achieved an accuracy of 96.3%, precision of 95.8%, recall of 96.7%, and F1-score of 96.2%, outperforming the CNN (93.5% accuracy) and SVM (88.4% accuracy) baselines. Hyperparameter tuning was shown to significantly enhance classification results, as visualized in the tuning heat map. The ROC curve for the hybrid model exhibited an Area Under the Curve (AUC) close to 1.0, indicating excellent sensitivity and specificity.