No Tags, No Stress: BelSU Unveils Smarter Farming

Researchers at Belgorod State University have developed a new “smart” system capable of recognizing farm animals with remarkable precision, identifying individual pigs in a herd with over 95% accuracy.

The hardware-software complex distinguishes each pig based on its unique facial biometrics, enabling contactless individual monitoring. The promising system is the result of a collaboration between scientists from Belgorod State University’s Institute of Engineering and Digital Technologies, led by Olga Ivashchuk, Head of the Department of Information and Robotic Systems, and colleagues from Sh. Yessenov Caspian State University of Technology and Engineering in Aktau, Kazakhstan.

According to the developers, the system stands apart from existing solutions by operating in an “open-enrollment” mode, meaning it can automatically detect a new animal within a herd and add it to its database without requiring retraining.

“The accuracy of animal detection in images is 99.5%, and the accuracy of individual identification exceeds 95%. These figures surpass those currently reported in the scientific literature,” Ivashchuk told RIA Novosti.

The expert emphasized that the new approach offers a less invasive alternative to widely used methods such as physical marking, sensor attachment, or ID tags – techniques that can be costly and stressful for the animals. The system continuously tracks movement as well as the frequency and duration of feeding, and could enable early detection of diseases. Initial testing of the system was conducted on pigs.

The technology operates in several stages: a neural network first detects animals within a video stream from cameras, and then matches each animal against a database. The system automatically monitors the feeding behaviour of every identified pig and alerts farm operators to any anomalies. Based on the data collected, it provides practical recommendations for farm staff, including suggestions for optimizing feeding schedules or adjusting enclosure temperatures.

As part of the ongoing project, the research team is also developing a mobile application that would allow farm personnel to identify pigs in real time using a smartphone camera.

The developers note that the system is highly scalable and can be adapted to farms of varying sizes. In the future, they aim to integrate it into digital management platforms not only for livestock farms but also for large-scale industrial agricultural complexes.