A Comprehensive Review of Complex Network Methods for Cotton Plant Disease Detection

Abstract

In recent years, researchers have become increasingly interested in cotton, a precious cash crop everywhere from Pakistan to China to Texas, is crucial for the textile industry and agricultural economies around the world. But its production is continuously threatened by a diversity of plant diseases, which can be triggered by pathogens such as bacteria, fungi and viruses. The conventional diagnostic approaches to confirm these diseases are generally not sensitive, time-consuming, and at the same time laborious. In recent times, network biology approaches have emerged as a robust platform in predicting disease associations and simulating diseases interplay in plants. The purpose of this paper is to investigate the application of complex network theory in cotton plant disease detection and control, particularly for the integration with machine learning and artificial intelligence. However, as many techniques are promising, only a few real-time interpretable field deployable models exist up to now to be used in local agricultural systems, especially in developing countries such as Pakistan. From this perspective, I propose that lightweight hybrid models should be developed which integrating network theory with spatiotemporal environment data, and open-access database would get higher availability. I believe there is a big opportunity in having a mobile-based decision support system, which is based on the graph learning model and can be used by farmers. This survey recommends that although substantial contributions have been achieved, there are still research opportunities for integrating complex networks into realistic and scalable systems regarding plant disease detection.

This paper aims to explore the application of complex network theory in cotton plant disease detection and management, focusing on integrating with machine learning and artificial intelligence. However, despite many promising methods, there is still a lack of real-time, interpretable, and field-deployable models that can be used in local agricultural systems, particularly in developing countries like Pakistan. Based on this analysis, I recommend the development of lightweight hybrid models that combine network theory with spatiotemporal environmental data, and the use of open-access datasets to enable broader adoption. I also see great potential in integrating mobile-based decision support systems powered by graph learning models for use by farmers directly. This review concludes that while significant progress has been made, the integration of complex networks into practical, scalable solutions for plant disease detection remains an open and valuable research direction.