PLANT DISEASE DETECTION USING ARTIFICIAL INTELLIGENCE: CURRENT TRENDS AND FUTURE PROSPECTS

Abstract

Plant diseases remain a big menace to agricultural productivity in the world, resulting into massive loss of produce, and compromised food security especially in areas experiencing resource shortages. Traditional methods of disease detection that are mostly reliant on expert visual evaluation are slow, subjective, and lack scalability. Over the past few years, the advent of the artificial intelligence (AI) and in particular, machine learning and deep learning algorithms have revolutionized the diagnostics of plant diseases by allowing the detection of the diseases through automated image-based methods at a high accuracy. Convolutional neural networks and computer vision technologies have demonstrated good potential to detect intricate patterns of diseases and facilitate precision farming. This review will critically analyze the advances in AI-driven plant disease detection to date, including the leading models, datasets, and performance trends. It was systematically and interpretively based on the methodological approach relying on the peer-reviewed literature of significant academic databases. It is emphasized in the analysis that deep learning models, and CNN-based architectures, in particular, and transfer learning training, in particular, are prevalent in the field and often achieve high accuracy when trained in controlled settings on benchmark datasets. Nevertheless, the review also notes the major limitations, such as the bias of the data set, low generalizability to real-life scenarios, high computational costs, and the absence of interpretability of the so-called black-box models. Such obstacles limit the applicability and adoption scale of AI solutions, particularly among the smallholder farmers. Moving forward, the evolution of multiple, field based datasets, explainable AI integration to enhance greater transparency, and edge computing to implement real time, on field diagnosis should be the focus of future research. It will also be necessary to strengthen the institutional support and digital infrastructure to close the technology development and practical application to achieve sustainable and inclusive agricultural transformation.

Plant phenotyping relevance

植物病害を画像から自動検出するAI手法を対象としたレビューであり、病害状態という植物表現型の取得・推定方法が中心です。

Code and data availability

This is a narrative review of AI-based plant disease detection. It contains no original phenotyping measurements, datasets, images, code, or models of its own; all cited datasets (e.g., PlantVillage) and DOIs belong to prior work, and no author code or data availability statement is present.

No evidence-backed public reproduction asset is currently recorded.