Media analysis for crop protection: Utilizing AI to monitor top five priority diseases in agriculture

Kim, Soonho; Song, Xingyi; Park, Boyeong; Ko, Daeun; and Liu, Yanyan. 2023. Media Analysis for Crop Protection: Utilizing AI to Monitor Top Five Priority Diseases in Agriculture. CGIAR

The report implemented under the CGIAR Initiative on Plant Health and details the development and implementation of a real-time media analysis system for assessing risks associated with the top 5 prioritized pests and diseases affecting crops cofounded by the Food Security Portal. This system, developed in collaboration with the University of Sheffield, utilizes a combination of text mining, machine learning techniques, and a Large Language Model (LLM), to process and analyze media articles. The goal is to identify patterns and assess the impact—quantitative and qualitative losses, as well as crop fatalities—caused by these pests and diseases. Throughout 2021-2022, the team tailored the media analysis system identified the most critical pests and diseases by the initiative. In 2023, the system was put into operation, and a cloud-based interface and REST API were developed to facilitate interaction with the analytical tools and integration with other systems. The interactive dashboard, which is publicly available, presents an interactive map and a detailed table displaying the outcomes of the media analysis. The system was evaluated and refined based on human verification, and manual corrections were fed back into the model for improvement. Looking ahead to 2024, the team plans to refine the system further, enhance the algorithm, and add more pests and diseases to the monitoring list. Monthly reports and updates to the CROP DISEASE DASHBOARD will continue to support policymakers and researchers in making informed decisions.