Deep Learning-Based CNN Framework for Automated Mango Leaf Disease Detection and Classification

Mango holds significant economic and cultural importance in Bangladesh. With mango production reaching 1.5 million tons in 2023, the country is ranked ninth globally in mango production. Pests and diseases, on the other hand, have had a significant impact on mango yields, which highlights the importance of finding effective disease management solutions. This study investigates the use of Deep Learning-Based convolutional neural networks (CNNs) for the automated detection and classification of mango leaf diseases, leveraging three models: VGG16, MobileNet, and DenseNet-121. Preprocessing was done to improve the quality of a large dataset of 4,352 images of mango leaves that covered a range of diseases. DenseNet-121 had the best accuracy (99%), followed by MobileNet (98%), and VGG16 (96%), according to our data. These models significantly improve the accuracy, reliability, and timeliness of disease detection, essential for early intervention and effective management. The study highlights the potential of advanced CNN architectures in agricultural applications, contributing to better crop health management, increased productivity, and enhanced food security. Future research will concentrate on growing the dataset and creating real-time applications to help farmers and agricultural specialists manage diseases.