Brazilian scientists have developed a platform to diagnose Asian soybean rust, one of the crop’s most damaging diseases. The technology combines artificial intelligence (AI) with integrated analysis of climate, agronomic, and digital image data. Hosted in the cloud, the system estimates the risk of disease occurrence and produces reports with technical management recommendations, supporting more precise on-farm decision-making.
The tool collects information from environmental sensors, digital images of soybean leaves, and agronomic parameters such as cultivar, row spacing, and sowing schedule. Results are displayed in an online dashboard that enables farmers to follow time series of climate conditions alongside plant images.
The platform was created under the Advanced Digital Tool for Agricultural Risk Management project, funded by the São Paulo State Research Foundation (Fapesp). The work was developed within the doctoral research of computer scientist Ricardo Alexandre Neves at the Federal University of São Carlos (UFSCar), supervised by Paulo Cruvinel, a researcher at Embrapa Instrumentation (São Paulo).
Findings from the project were published in July 2025 in the journal AgriEngineering in the article “A Cloud-Based Intelligence System for Asian Soybean Rust Risk Analysis in Soybean Crops.”
Data fusion supports diagnosis
The system was built through on-farm research using a model that merges climate variables, soybean crop data, and information extracted from digital images of soybean leaves. Climate conditions were monitored throughout the field observation period.
“The technology classifies disease favorability into three levels — low, medium, and high — depending on the combination of variables that relate to the stage of infestation. That allows diagnoses and prognoses for disease control with higher effectiveness and accuracy,” Neves adds. According to him, the level of favourability is defined by statistical inference based on the behavior of the set of variables.
The researchers explain that the system operates through data integration. Key inputs allow the analysis of conditions critical to fungal development, such as the leaf-wetness period — defined by relative humidity above 90% within a temperature range of 15°C to 28°C — and the dew point.
The work applies advanced, specialized processing techniques to extract information from digital images of soybean leaves. Color patterns — such as green, yellow, and brown — are used to indicate stages of disease progression.
Cruvinel notes that two approaches were assessed to combine the different data streams. Ultimately, the system adopted a Hidden Markov Chains model, selected for its robustness, effectiveness, and efficiency in supporting decisions. In the study, this method outperformed fuzzy logic, reaching 100% accuracy in matching the evaluated scenarios for Asian soybean rust risk in monitored fields.
“The model that was developed to merge data on different variables made it possible to structure a complete set of rules that systematically considers different situations in which the disease is likely to occur,” the researcher says.
During the four-year study with Embrapa Soybean’s conventional cultivar BRS 536, researchers used more than 2 gigabytes of data per crop cycle, considering information collected in actual fields during cultivation, in georeferenced plots in the Poxoréu-MT region and photographed under known lighting indices, according to a press release.
Data available to farmers online
The dashboard’s analytical reports were developed using two decades of historical data, allowing users to evaluate key periods across the crop cycle. The interface is designed for easy navigation and is organized around basic information relevant to farmers and other potential users.
According to Cruvinel and Neves, the reports are intended to support decisions on crop-area management by helping assess whether Asian soybean rust is present, as well as the severity of infection. They also provide agronomic recommendations for disease control based on the diagnostic results.
Cruvinel notes that the reports are accessible through the “Agricultural Recommendations” tab on the dashboard. This section also includes a link to the AGROFIT website, a database maintained by Brazil’s Ministry of Agriculture and Livestock (Mapa) with information on registered agrochemicals and related products, supporting consultation and selection of fungicides recommended for Asian rust management.
Solution reduces fungicide use
The researchers add that the system enables monitoring of Asian soybean rust presence or absence and supports evaluation of disease dynamics across different levels of severity and risk throughout the production process.
“The key point of the research was to create a method that integrates heterogeneous data to provide a more reliable diagnosis. Relying solely on images or isolated climate data is not sufficient for an accurate assessment, which can lead to false-positive diagnoses. Besides that, the solution offers prevention and rational use of fungicides,” says Neves, who is currently a professor at the Federal Institute of São Paulo (IFSP), São João da Boa Vista campus.
