Optimizing Citrus Disease Prediction: A Hybrid CNN-SVM Approach for Enhanced Accuracy

Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) were expertly combined in this study to create a powerful hybrid model for the automated diagnosis of citrus diseases. This model was developed as a result of this research. The model, which is comprised of three convolutional layers, three max-pooling layers, and two fully connected layers, displays remarkable performance across a variety of citrus illnesses, including Citrus Canker, Citrus Greening, Melanose, Citrus Scab, Alternaria Brown Spot, and Phytophthora Root Rot. In addition, the model is characterized by its three convolutional layers. The analysis of the layer parameters highlights the significance of the selected architecture in terms of optimizing predictive capabilities. The hybrid strategy, which achieves this result by methodically utilizing CNN for the purpose of feature extraction and SVM for the purpose of classification, obtains an accuracy of 92.34% overall. The execution of the technique in stages provides a methodical and exhaustive process of model development, which ultimately results in improved disease prediction accuracy. Beyond the citrus orchard, this study makes a contribution to the landscape of precision agriculture by arguing for the incorporation of cutting-edge machine learning models into contemporary agricultural methods. This research serves as a model for harnessing innovative technologies to boost production and build sustainable disease management techniques, which is particularly useful at a time when the agricultural industry is facing new and evolving issues.