Noncontact Weighing of Cattle and Sheep Using AI to Predict Growth
Researchers from Chelyabinsk State University and the Federal Scientific Center for Biological Systems and Agrotechnologies of the Russian Academy of Sciences have created a noncontact weighing method for cattle and sheep. The system uses artificial intelligence to estimate weight and predict future growth when a camera is pointed at the animal. The findings were shared with socialbites.ca and attributed to the Russian Ministry of Education and Science.
Today, external assessment of livestock in industrial farms relies on manual expertise, a process that can be subjective. Traditional weighing also stresses the animal and can lead to weight loss. The new method avoids these drawbacks while delivering reliable measurements.
By employing computer vision, the system captures data from images to estimate body weight. This approach also reveals morphological traits and supports forecasting weight changes over time, enabling better management decisions.
The research lead, Aleksey Ruchai, who heads the Computer Security Department and the Applied Algebra section at ChelGU, explained that the system determines body weight without contact and helps categorize meat quality. He noted that predictions about marbling can guide decisions on slaughter timing, ultimately lowering livestock production costs. Source: Russian Ministry of Education and Science.
At present, the team is refining the underlying database, a key driver of the model’s accuracy and efficiency. Experiments continue with noncontact weighing across different animal groups to expand the dataset. Initial measurements showed an error margin of five to ten percent, a performance level comparable to traditional scales.
The project began with three rooms for camera calibration and capturing animals in a still position. The researchers are now testing a single-camera setup as animals pass by, aiming to deliver a simple, versatile, fully automatic weighing solution that can be installed without specialized training.