Computer Vision Reveals Social Networks in Dairy Cattle: Valence-Aware Analysis

It uses computer vision to classify animal interactions as positive (affiliative) or negative (agonistic) to build social networks revealing group organization.

It can be integrated into smart monitoring systems to improve animal welfare, reduce stress, and increase milk yield, aligning with food security goals.

Automated-manual agreement was 82.8% in an audited sample, with overall macro-F1 of 0.872, but full deployment requires longitudinal validation.

Not yet; it needs longitudinal validation before use as a welfare indicator, but it provides a strong foundation for development.