Computer vision study tracks pig behavior in climate-controlled barns

Pigs housed in a temperature-controlled barn could be monitored for feeding, drinking, standing, and lying behavior using a computer-vision system built on YOLOv5, according to a study published April 28, 2026, in Animals. Researchers installed microcameras in stalls for 92 days, paired the video with temperature and humidity data, and also collected physiologic measures to assess heat stress. The team reported that the system achieved more than 97% accuracy for detecting pigs and recognizing feeding and drinking behaviors, with the goal of replacing labor-intensive manual observation in an air-conditioned environment. (mdpi.com)

Why it matters: For veterinary professionals working in swine production, the study adds to a growing body of precision livestock farming research suggesting cameras can support earlier, less intrusive monitoring of welfare-relevant behaviors. Prior reviews have linked camera-based monitoring to detection of feeding, drinking, lying, and locomotion patterns, and earlier work in Scientific Reports showed automated posture and drinking recognition could detect routine disruptions with very high precision. At the same time, experts caution that camera systems still face real-world barriers, including lighting variation, occlusion, and the gap between university validation and commercial-farm deployment. (mdpi.com)

What to watch: The next question is whether systems like this can hold up in commercial barns, integrate with climate-control and health data, and move from behavior description to earlier detection of disease, stress, or welfare problems at scale. (mdpi.com)