Jadwal Sholat

Memuat jadwal sholatโ€ฆ

Pertanian & Lingkungan editorial

Open AccessOA2024

Revolutionizing agriculture with artificial intelligence: plant disease detection methods, applications, and their limitations

This review examines AI and IoT-based methods for plant disease detection in four key vegetable crops, detailing a five-step pipeline from image acquisition to classification. It evaluates ML and DL models, discusses datasets, and identifies limitations and future research directions.
Abbas Jafar; Nabila Bibi; R. Naqvi; Abolghasem Sadeghi-Niaraki; Daesik Jeongยท Frontiers in Plant Scienceยท 2024ยท DOI 10.3389/fpls.2024.1356260

The core problem

Plant diseases caused by viruses, fungi, bacteria, and other pathogens pose a major threat to agricultural productivity, leading to significant economic losses. Traditional detection methods are time-consuming, require expert knowledge, and are resource-intensive. The authors argue that automated leaf disease diagnosis using artificial intelligence (AI) combined with Internet of Things (IoT) sensors can provide faster and more accurate detection. This review focuses on four economically important vegetable crops: tomato, chilli, potato, and cucumber. It highlights the most prevalent diseases and infections in these crops, along with their symptoms, and provides a structured approach to predicting plant diseases using AI. The study aims to survey existing machine learning (ML) and deep learning (DL) models, discuss datasets, and identify application challenges and future research prospects, particularly the integration of AI with IoT platforms such as smart drones for field-based monitoring.

Innovation

The review outlines a predetermined pipeline for AI-based plant disease detection consisting of five key steps: image acquisition, preprocessing, segmentation, feature selection, and classification. Image acquisition involves capturing leaf images using smartphones, drones, or other IoT devices. Preprocessing techniques such as resizing, filtering, and normalization are applied to enhance image quality. Segmentation separates the diseased region from the healthy part of the leaf. Feature selection identifies relevant visual features (e.g., color, texture, shape) that are indicative of disease. Finally, classification uses ML or DL models to categorize the disease. The authors discuss both traditional ML algorithms (e.g., support vector machines, k-nearest neighbors, random forests) and DL architectures (e.g., convolutional neural networks). They also compile a list of publicly available datasets used for evaluating these models, such as PlantVillage, and provide a comprehensive examination of existing studies for the four crops.
Introduction
Plant diseases caused by viruses, fungi, bacteria, and other pathogens pose a major threat to agricultural productivity, leading to significant economic losses. Traditional detection methods are time-consuming, require expert knowledge, and are resource-intensive. The authors argue that automated leaf disease diagnosis using artificial intelligence (AI) combined with Internet of Things (IoT) sensors can provide faster and more accurate detection. This review focuses on four economically important vegetable crops: tomato, chilli, potato, and cucumber. It highlights the most prevalent diseases and infections in these crops, along with their symptoms, and provides a structured approach to predicting plant diseases using AI. The study aims to survey existing machine learning (ML) and deep learning (DL) models, discuss datasets, and identify application challenges and future research prospects, particularly the integration of AI with IoT platforms such as smart drones for field-based monitoring.

Why it matters

The authors critically analyze the limitations of current AI-based plant disease detection systems. Key challenges include the lack of large, diverse, and annotated datasets that represent real field conditions; the difficulty of generalizing models across different crops, diseases, and environmental conditions; and the computational demands of DL models, which may hinder deployment on resource-constrained IoT devices. Additionally, the integration of AI with IoT platforms like smart drones is still in its infancy, with issues related to data transmission, power consumption, and real-time processing. The review also discusses the importance of early detection to minimize yield losses and the potential of combining multiple data sources (e.g., visual, spectral, environmental) to improve accuracy. Future research directions include developing lightweight DL models for edge computing, creating standardized benchmarks, and conducting more field trials. The authors conclude that while AI has revolutionized plant disease detection, significant work remains to translate laboratory success into practical, scalable solutions for farmers.

Who should read this

Pembaca Research Digest dan praktisi rumpun ini.

Opening member contentโ€ฆ