Russian scientists move towards climate-resilient agribusiness

Photo by: Jenson / iStock

Scientists at Tomsk State University (TSU) have developed a neural network capable of analysing the condition and fertility of agricultural land using satellite imagery. This was reported by the university’s press service.

The project was implemented with the support of the St. Petersburg-based IT company Syncretis, which specializes in creating systems for recognition and classification of various objects using machine vision and artificial intelligence (AI).

“Precision farming tools are particularly needed in Siberia and other areas that are risky farming areas. The use of neural networks will help to reduce losses and increase crop yields. TSU biologists together with the university’s partner, IT company Syncretis, taught AI to analyse field fertility and crop condition using satellite imagery. The new tool will be available to Russian agronomists developing precision farming technologies”, — the university said.

According to representatives of the university, the scientists and programmers’ development is unique for Russia and has been patented.

As the biologists specified, the AI recognises different types of soil using biomarkers — a system of special marks created especially for the project. They are used by the neural network to read all of the most important functional features of a particular area of the field.

It took the scientists two years to create the library of biomarkers. TSU also developed sensors to analyse soil and surface air indicators.

Next, soil scientists investigated agricultural land, took soil samples and determined patterns between the reflectivity of land with crops and how they look on satellite images.

The final part of the project was to train a neural network on the array of data collected.

As a result, the AI determines the level of fertility of particular fields from orbital photographs, identifies possible areas of crop damage, and calculates the causes of the damage. The virtual system can also suggest options for identifying a problem in more detail and fixing it, such as sending drones to scout or treating the soil with the right preparations.

Next year, TSU intends to test the AI on the fields of its industrial partner in the Novosibirsk Region. The developers are now deciding on a platform to host the service, making it available to Russian farmers and agricultural enterprises.

Cover photo: Jenson / iStock