Malaysian Study Harnesses AI to Transform Citrus Disease Detection

88

• A Malaysian study has developed an AI-powered system capable of detecting and classifying citrus leaf diseases with high accuracy, offering farmers a faster and more reliable diagnostic tool.
• The technology uses image recognition and machine learning algorithms to identify early signs of infection, helping growers take timely action before crop losses escalate.
• By reducing reliance on manual inspections and expert intervention, the AI approach lowers costs while increasing efficiency in citrus production.
• Researchers emphasise that this breakthrough could strengthen global citrus supply chains, ensuring healthier yields and improved food security.

A pioneering study led by Bobbinpreet Kaur from Lincoln University College, Malaysia, is set to transform how citrus diseases are identified and managed. 

Published in the respected journal Scientific Reports, the research introduces a cutting-edge automated system capable of detecting and classifying citrus leaf diseases with exceptional accuracy. The development promises significant benefits for the global citrus industry, which underpins both food security and economic growth.

Citrus fruits, especially lemons, play a crucial role in global agriculture. However, they are increasingly vulnerable to a range of diseases that can severely affect both yield and quality. 

Traditional detection methods whether manual or automated are often labour intensive, require expert knowledge, and struggle to detect infections at early stages. Kaur’s research directly tackles these limitations through a blend of image enhancement techniques and advanced deep learning models.

(Read Also: FarmLab Acquires Ziltek to Expand Soil-Scanning Tech Beyond Agriculture)

The process begins with image optimisation using Vector-Valued Anisotropic Diffusion (VAD) and morphological filtering. 

These methods enhance the clarity of citrus leaf images, a vital step for reliable classification. “The clarity of the images is paramount,” Kaur explains. “Without it, even the most advanced models may fail to accurately detect disease.”

Central to the system is the Nonlinear Fuzzy Rank-Based Ensemble (NL-FuRBE) methodology. This approach integrates three powerful deep learning architectures: VGG19, AlexNet, and Xception within a fuzzy rank-based scoring framework. 

Nonlinear transformations, including exponential, tanh, and sigmoid functions, are then applied to refine predictions, reduce bias, and improve overall reliability. “By combining these models and transformations, we achieve a level of accuracy previously thought unattainable,” Kaur adds.

This innovation marks a major step forward in precision agriculture. By enabling faster, more accurate disease detection, it offers growers a powerful tool to protect citrus crops, improve yields, and strengthen global food security.