AI System Identifies Pigs by Their Snouts, Russian Researchers Say
The smart monitoring technology is designed for use at pig farms.
Researchers at Belgorod State National Research University have developed a system that can recognize individual pigs within a herd. The project was created jointly with colleagues from the Yessenov Caspian State University of Technologies and Engineering.
The hardware–software complex is based on a mechanism that distinguishes animals by their unique traits. Pigs are identified using biometric parameters of their snouts through a neural network operating in an open-set recognition mode. While closed-set systems recognize only objects included in the training dataset, open-set recognition automatically incorporates newly detected animals.
Monitoring Animal Health
The monitoring system can track animals remotely and without physical contact, following their movement paths and analyzing how often and how long they feed. This data can help detect diseases at early stages.
The system operates in several stages. First, the neural network detects animals in video streams from cameras, then identifies them using a database. After that, the system automatically tracks the feeding behavior of each individual and alerts operators about deviations. It can also suggest optimal feeding times and recommend appropriate temperature conditions in the enclosure.
In the future, researchers plan to develop a mobile application that will allow animals to be identified in real time using a smartphone camera.
