A COMPARATIVE ANALYSIS OF DEEP LEARNING MODELS FOR MANGO LEAF DISEASE DETECTION AND CLASSIFICATION

Mango (Mangiferaindica) being an economically and nutritionally significant tropical fruit, yet its cultivation is threatened by several foliar diseases such as anthracnose, powdery mildew, bacterial canker and so on. However, this is an automatic identification for mango plant disease and classification has vied an important role within agriculture exploitation digital image process techniques. Basically, traditional detection methods rely on manual inspection, which is labor-intensive, subjective, and often delayed. Advances in deep learning (DL) provide opportunities for automated, accurate, and scalable solutions. This study presents a comparative analysis of deep learning models for mango leaf disease classification using a dataset of 4,000 images across seven classes: healthy, anthracnose, powdery mildew, bacterial canker, gall midge, dieback, cutting weevil, and sooty mold. Four models namely: Custom CNN, LeafNet, AlexNet, and VGG19, were trained and evaluated using accuracy, precision, recall, and F1-score. Results show that Custom CNN and LeafNet achieved the highest performance (99.5% across all metrics), followed by VGG19 (99.0%) and AlexNet (88.0%). The study also introduces a vein-pattern-based segmentation approach that enhances feature localization. The findings highlight the potential of AI-driven frameworks for early mango disease detection, with implications for improving crop management, reducing yield losses, and supporting sustainable agricultural practices.