What you’re seeing
Real: Bounding boxes, pig tracking IDs, and confidence scores come from real YOLOv8 inference (Open Images V7 “Pig” class) on stock barn footage, with ByteTrack maintaining IDs across frames.
Derived: Weight estimates, activity states (lying / standing / locomotion), velocity, and feeding-zone dwell are computed from the detection data using research-grade formulas documented in the build pipeline.
Illustrative: 7-day metric trends, the social hierarchy graph, and 9 of the 12 pens in the barn overview use deterministic synthetic data so the demo shows a fully populated barn with only 4 source clips.
ConductVision is technology available for pilot deployment, not commercially deployed telemetry. Talk to us about your operation →