Diagnosis of Mango Leaf Diseases Using Deep Learning Techniques

Mango cultivation is a cornerstone of Thailand's agriculture and economy, but diseases such as anthracnose, algal leaf spot, and gall midge present significant challenges as they can reduce crop yield and quality. In this study, we developed a machine learning-based system to diagnose mango leaf diseases using a dataset of 1,900 images collected from mango orchards in Phitsanulok province. The data underwent preprocessing and augmentation to optimize model training. Five deep learning models—Convolutional Neural Network (CNN), VGG16, DenseNet121, ResNet50, and InceptionV3—were trained and evaluated. Among these, ResNet50 demonstrated the best performance, with an accuracy of 99.8%, a precision of 0.998, a recall of 0.998, and an F1-score of 0.998. Leveraging its superior performance, the ResNet50 model was integrated into a mobile application designed for real-time disease diagnosis. This user-friendly application enables mango farmers to upload images of affected leaves and receive instant disease identification and treatment recommendations. The findings highlight the potential of deep learning models in agricultural applications, offering a reliable and efficient tool for early disease detection and management. By enabling timely intervention, this innovation enhances crop health, reduces losses, and boosts productivity, contributing significantly to sustainable farming practices and improving farmers' livelihoods.