AI-driven banana pest and disease management: methods, applications, challenges, and future directions
摘要
Banana production faces escalating threats from pests and diseases, especially in smallholder systems. This review provides a comprehensive assessment of artificial intelligence applications in banana health monitoring, focusing on deep learning models including convolutional neural networks, YOLO variants, and Vision Transformers. Unlike prior reviews, this work classifies model performance across mobile, UAV, and IoT-based systems and evaluates effectiveness under both controlled and field conditions. It introduces recent advances such as Swin Transformer and lightweight ViT architectures. Beyond technical capabilities, the review highlights barriers related to data quality, model generalization, cost, and infrastructure, and proposes strategies for improving adoption. By bridging technical advances with policy insights, the review offers a structured roadmap for scalable, inclusive AI deployment in banana pest and disease management.