Multi-Crop Multi-Disease Detection Framework Using Explainable Artificial Intelligence for Precision Agriculture: A Comprehensive Literature Review

Abstract

Plant diseases pose significant menace to food security in the world, causing significant losses of money and reduced yields of key agricultural products like cotton, tomato, wheat and rice. The traditional methods of detection that rely on manual diagnosis are time consuming, labour-intensive and prone to human errors hence the necessity of automated and accurate diagnostic systems. The recent advances in artificial intelligence, especially deep learning and computer-vision methods, have changed the scene of the plant disease detection because it has made it possible to identify various diseases on different crop species in real time, automatically, and accurately. In this literature review, we discuss the state of the art in multi-crop disease detection models, focusing on deep-learning models, such as Convolutional Neural Networks (CNNs), Vision Transformers (ViT), and object-detection models, including YOLO, Faster-R-CNN, and Mask-R-CNN, and interpretable AI (XAI) methods, such as Grad-CAM, SHAP, and LIME. Through a systematic review of thirty recent articles released in 2020-2025, the review outlines major technological developments, performance standards, as well as viable deployment plans of precision agriculture. It also determines serious gaps in research such as lack of integrated multi-crop models, little verification of these in field applications, inability to scale computations to support edge deployment, and lack of model interpretability. The review thus adds to the field through compiling the existing knowledge, providing a comparative methodology analysis, and future research steps to create explainable, lightweight, and farmer-friendly AI systems to achieve sustainable agriculture.