ST. PETERSBURG, June 19. /TASS/. Scientists at the St. Petersburg Federal Research Center (the Russian Academy of Sciences) created Russia's first automated neural network system capable of analyzing reindeer external features to assess their conditions by using computer vision. The development will solve a range of problems that the livestock industry in the Arctic is facing, the Russian Ministry of Science and Higher Education told TASS.

"The reindeer industry is facing a number of challenges, including a shortage of qualified specialists at farms, as well as insufficient data on heredity," the ministry said. "One of solutions to this problem is to develop automation tools that will increase the productivity of industry employees, as well as increase the quantity and quality of information about reindeer."

According to Vladislav Sobolevsky, a senior researcher at the center's Laboratory of Information Technologies in System Analysis and Modeling, the development is based on a neural network that uses computer vision to analyze the phenotypic data of animals on videos. This way, scientists can analyze the condition of every animal. "In addition, our system monitors deer without contact and thus eliminates stress factors that occur in the animal, for example, when using wearable sensors," he said.

While training the neural network, specialists picked 60 photographs of reindeer made in the Yamal experimental herd. In experiments, the system demonstrated the ability to recognize accurately all key biometric parameters that allow assessing an individual's health status.

"Monitoring of deer fatness by analyzing muscle mass makes it possible to predict accurately the herds' productivity, which is critical for managing the food supply in conditions of limited pasture resources. And computer detection of pathologies by the skin texture is used for early diagnosis of diseases, thus cutting death losses," the ministry quoted the center's head of the Animal Husbandry and Environmental Management Department, Kasim Laishev, as saying.