Mango Disease Detection System Using CNN and SVM: A Comparative Study with Traditional Diagnostic Methods

This study proposes a hybrid deep learning approach to classify diseases into nine diseases by combining support vector machines and convolutional neural network models. In classifying various diseases in banana fruit leaves, the research presents a hybrid deep convolutional neural network (CNN) and support vector machine (SVM) algorithm. 1570 images covering 9 distinct disease groups help compensate for the dataset used in the study. The study’s CNN architecture, which includes several convolutional and max pooling layers, fully connected layers, and an SVM output layer in which we implement regularisation, is also thoroughly discussed in the research. Based on the study results, the proposed CNN-SVM model achieves an overall accuracy of 81.13%, with macro average values for accuracy, recall, and F1 score of 81.8%, 82.66%, and 81.85%, respectively, and weighted average values of 40.80%, 41.60%, and 41.19%. These results indicate how machine learning algorithms can be applied in agriculture to quickly and precisely diagnose disease, which can ultimately improve crop management and production.