Farm Welfare Score
Alert Summary
Key Metrics
3.1
Avg BCS
20%
Lameness %
10.8
Avg Lying (hrs)
238
Avg Feeding (min)
Total Cows
50
Active Zones
5
Alerts Pending
4
Avg Rumination (min)
486
Real ConductVision Output — Run 1: Detection and Tracking
10 cows · 73.2% confidence · 78s clipEvery cow is detected and kept on the same track across the whole clip, using ConductVision's production detection model with no cattle-specific retraining.
Public dairy-barn footage, 640 × 360, 30 frames per second · 2,338 frames processed · 6.7 ms per frame
Real ConductVision Output — Run 2: Body Outline and Walking-Gait Proxy
6 cows segmented · 4 with gait numbers · no retrainingA second pass on the same footage extracts each cow's body outline (silhouette) and pulls a normalised back-arch measurement from it. Back arch is a clinically validated lameness indicator (Sprecher 1997; Flower & Weary 2006). No cattle-specific retraining was done; this is an early proof of the pathway, not a calibrated clinical score.
| Cow | Frames | Back arch (0–1) | Head bob (px) | Stride regularity |
|---|---|---|---|---|
| Cow 0 | 221 | 0.078 | 31.8 | 1.88 |
| Cow 1 | 120 | 0.048 | 36.9 | 1.51 |
| Cow 2 | 10 | 0.045 | 60.9 | 0.68 |
| Cow 3 | 5 | 0.046 | 59.8 | 0.68 |
Back arch is normalised on a 0 to 1 scale (higher = more arched). Head bob is vertical centroid motion in pixels. Stride regularity is how much step length varies between steps; lower is more regular. Numbers trace to results/gait/metrics.csv in the submission repo.
System Comparison
Architectural approaches to cattle welfare monitoring
| Metric | Choke-Point Systems | Full-Facility Systems |
|---|---|---|
| Observation coverage | —Single point (e.g., parlor exit) | ✓Pens, bunks, lanes, water, pasture |
| Behavioral metrics | —Mobility focus | ✓Full ethogram (6+ categories) |
| Video processing | —Cloud | ✓Edge (on-farm) |
| Facility coverage | —One location | ✓Entire operation |
| Operation types | —Dairy-focused | ✓Dairy, beef, cow-calf |
| Intelligence layer | —Reactive alerts | ✓Digital Twin + scenario modeling |
30-Day Herd Welfare Trend
Recent Alerts
View allBack arch asymmetry elevated (+38% vs baseline). NRS progressed 2→3 over 5 days.
Schedule locomotion assessment. Consider hoof trimming.
Lying time -15%, restlessness +40% vs 14-day baseline. Early mastitis pattern detected.
Urgent: veterinary examination for clinical mastitis signs.
HLI forecast 89 for Zone: Holding Area in 6 hours. Threshold: 86.
Activate sprinklers in holding area. Delay afternoon milking shift by 1 hour.
BCS declined 3.2→2.7 over 30 days. Below optimal range.
Review feed intake. Check for subclinical health issues.