Biologists and mathematicians of Altai State University have developed a plant monitoring system based on artificial intelligence — TASS

Scientists from Altai State University, together with Yandex Cloud, have developed a plant monitoring system that, using a neural network, will help predict the spread of pests and prevent crop losses among farmers. It is reported by TASS with reference to the press service of Yandex Cloud.

The machine learning algorithm predicts the distribution of plants in the territory specified by the user in the system. In addition to scientific monitoring, the system is able to help solve the problems of agricultural producers. Due to the infestation of crops in Russia, about 17% of the grain harvest is lost annually. With the help of the new system, farmers will be able to more accurately predict the distribution of pests of each individual species. This will determine the best locations and exact area for crops, as well as plan harvesting and tillage methods.

“AltSU botanists started working with big data on biodiversity only recently, but have already received recognition as development support from the Russian Science Foundation and the Potanin Foundation. Finally, in 2022, together with the mathematicians of Altai State University, we started working on a serious system that allows us to predict the distribution of plants in a certain area. The impetus for this initiative was the Priority 2030 program of the Ministry of Education and Science of the Russian Federation. We went beyond the usual framework and, together with the strategic partner Yandex.Cloud, initiated the creation of a completely new product for the country - a virtual biolaboratory for the analysis and assessment of biodiversity and climate modeling. The work ahead is difficult and it will not be solved immediately, but the first results of the biolaboratory have already pleasantly surprised us,” said Alexei Vaganov, researcher at SSBG and head of the AltSU strategic project.

The system is deployed on the Yandex Cloud cloud platform and predicts plant distribution based on several types of data. This is information about the points of presence of this species, climate, humidity and soil type. The user can enter the name of the plant in Latin into the system, upload the dataset and, after processing the request by neural networks, get graphs and detailed maps with a forecast.

The generated heat map allows for assessment in which areas the plant can be found in the future. Now the tool is working in an experimental mode and is being tested on plant species of Altai Krai. In the future, the system will make forecasts in other regions of Russia. It is planned to improve the tool with the help of computer vision. The user will be able not to enter the type of plant into the system, but to take it on a smartphone camera and upload a photo. Using the image recognition service, the system will automatically identify each view.

The development, according to scientists, may be in demand for the markets of agro-bioindustry, health care and environmental protection.