Systematic review of machine learning models for plant and leaf disease detection and classification

Plant leaf diseases significantly affect crop yield and quality, posing ongoing challenges across the agriculture sectors, including cereals, vegetables, fruits, and floriculture. In response, a wide range of machine learning (ML) techniques have been explored for effective disease detection and classification. This systematic review provides a comprehensive analysis of 102 peer-reviewed studies published between 2001 and 2024, exclusively focused on ML-based approaches for identifying plant leaf diseases. Based on the PRISMA methodology, studies were selected from four major databases and organized under eight research questions (RQs), covering the end-to-end ML pipeline. The review investigates key components such as dataset collection methods, model architectures, evaluation strategies, and comparative model performance. It further analyzes the distribution of studies by plant type, disease category, and publication source. Special emphasis is placed on dataset characteristics, model generalization across domains, and challenges in real-world implementation. The review also highlights performance metrics, cross-validation strategies, and the need for scalable, explainable, and lightweight models. Gaps in the literature—such as limited availability of high-quality datasets, lack of standard benchmarking, and poor model transferability—are identified to guide future research. This focused review offers a structured, comparative overview of ML-driven leaf disease classification efforts and serves as a roadmap for researchers and practitioners aiming to enhance agricultural productivity through intelligent, data-driven plant health diagnostics.