A Comparative Analysis of Machine Learning and Deep Learning for Detecting and Classifying Mango Leaf Diseases

Abstract

The rapid advancement of machine learning (ML) and deep learning (DL) techniques has opened up new avenues for automated disease detection and classification in agricultural settings. This paper presents a comprehensive comparative analysis of ML and DL algorithms for detecting and classifying mango leaf diseases. The ML algorithms employed in this study include K-nearest neighbors (KNN), support vector machines (SVM), and random forest (RF), while the DL models evaluated are convolutional neural network (CNN), VGG16, and Inception V3. To conduct this analysis, a dataset com- prising a diverse range of mango leaf images infected with different diseases was collected and preprocessed. The ML algorithms were trained using features extracted from the images, whereas the DL models were trained end-to-end using the raw image data. The experimental results revealed that DL models achieved higher accuracy and robustness compared to the ML algorithms. The CNN, VGG16, and Inception V3 models demonstrated superior performance in terms of disease detection and classification, but the accuracy levels achieved by KNN, SVM, and RF algorithms are not as good as DL models. Moreover, the DL models exhibited remarkable capability in capturing complex patterns and intricate features in mango leaf images, leading to more accurate disease identification. In conclusion, this study provides a comprehensive comparative analysis of ML and DL algorithms for mango leaf disease detection and classification. The findings highlight the advantages and limitations of each approach, allowing researchers and prac- titioners to make informed decisions when selecting appropriate techniques for similar agricultural applications. The results underscore the potential of DL models.