Document Type : Review Papers
Authors
1 Department of Agricultural Engineering, Afghanistan National Agricultural Sciences and Technology University,(ANASTU).
2 Department of Horticulture, Afghanistan National Agricultural Sciences and Technology University (ANASTU), Kandahar, Afghanistan.
3 Department of Plant Protection, Afghanistan National Agricultural Sciences and Technology University (ANASTU), Kandahar, Afghanistan
4 Department of Agronomy, Afghanistan National Agricultural Sciences and Technology University (ANASTU), Kandahar, Afghanistan
Abstract
Artificial Intelligence (AI) and Machine Learning (ML) have emerged as important technologies for advancing precision agriculture through data-driven decision-making, real-time monitoring, and automation across the crop production cycle. This review provides a comprehensive assessment of AI-driven agricultural technologies and their integration with enabling systems, including Unmanned Aerial Systems (UAS), remote sensing, the Internet of Things (IoT), big data analytics, and robotics. Unlike previous reviews that primarily focus on specific AI techniques or individual agricultural applications, this review offers an integrated perspective on technological architectures, practical applications, implementation challenges, and sustainability implications within the agriculture 4.0 framework. AI-based approaches support crop health monitoring, disease and pest detection, resource management, and crop yield prediction through advanced machine learning and deep learning techniques. In particular, Convolutional Neural Networks (CNNs), Vision Transformers, and hybrid architectures have demonstrated strong capabilities in processing multispectral, hyperspectral, and multimodal agricultural data. Despite their potential to improve productivity, resource-use efficiency, and environmental sustainability, several challenges continue to limit widespread adoption, including data scarcity, limited model generalizability, interoperability constraints, high implementation costs, and insufficient digital skills among smallholder farmers. Additional concerns related to data privacy, ownership, transparency, and algorithmic bias further complicate implementation. This review discusses recent advances, current limitations, and future research directions, emphasizing the development of scalable, interpretable, and human-centered AI systems, as well as the integration of federated learning and physics-informed approaches to support resilient and sustainable agricultural systems.
Keywords
)