Image-Based Disease Detection in Agriculture: Advances, Challenges, and Applications in Oil Palm Management

Abstract

Image-based disease detection plays a key role in precision agriculture. It has advanced rapidly with developments in computer vision, deep learning and sensing technologies. This review summarizes recent research on agricultural disease detection, with a focus on oil palm management. The main imaging modalities are reviewed, and traditional computer vision methods are compared with state-of-the-art deep learning and transformer-based models. Their performance across different agricultural conditions is discussed. Applications in oil palm cultivation are highlighted. This includes early detection of Ganoderma basal stem rot, identification of nutrient deficiencies, and farm-scale monitoring. Although detection accuracy and field applicability have improved, several challenges remain. These include limited annotated datasets, environmental variability, high sensor cost and poor model generalization. Future research directions are outlined, emphasizing multimodal sensing, use of edge AI, data set expansion and integration with digital farm management systems. Overall, image-based detection technologies offer strong potential to improve crop monitoring, support decision-making and enhance the sustainability of oil palm production.