An investigation on advances in transfer learning and explainable AI for mango leaf disease detection

Mangoes (Mangifera indica) are one of the most economically significant tropical fruits, but various leaf diseases increasingly threaten their cultivation. Early detection of these diseases is critical to preventing crop loss and ensuring sustainable agricultural practices. Traditional disease detection methods, reliant on expert observation and manual inspection, are time-consuming and prone to human error. This paper investigates an Artificial Intelligence (AI)-driven approach for the automated detection of mango leaf diseases using deep learning models, specifically Convolutional Neural Networks (CNNs) and Vision Transformers (ViT). We utilize transfer learning with pre-trained models on ImageNet, including ResNet50, DenseNet121, Inception V3, EfficientNet B0, VGG 19, ViT-B-16, ConvNeXt, Swin Transformer, and Data-Efficient Image Transformer to improve performance on a dataset consisting of mango leaves images, classified into several disease categories. The dataset is split into training, validation, and test sets to train and evaluate the models using various performance metrics. The result analysis demonstrates that Swin transformer scores well in accuracy, precision, F1 score, and recall compared to other models, proving their exceptional capability in disease classification. Furthermore, the paper integrates Explainable AI (XAI) techniques such as Gradient-Weighted Class Activation Mapping (Grad-CAM) and Local Interpretable Model-Agnostic Explanations (LIME), to provide visual insights into the model's decision-making process, enhancing model interpretability and trust. The results demonstrate the potential of these deep learning models for accurate disease classification, and the application of Explainable AI ensures that the detection process is transparent and understandable. This investigation contributes to more efficient, scalable, and reliable disease management in mango cultivation, offering a step towards sustainable agriculture powered by AI.