Leveraging computer vision to identify animals at risk of injury

A scalable, real-time injury detection system using computer vision

OneSix built a horse workout tracking application that leverages an overhead multi-camera array, on-track cameras, and an ultra-wideband tracking system to monitor and derive critical metrics from workouts.

• Installed 7–10 trackside cameras across three unique tracks, along with a complete overhead camera array.
• Developed a robust production pipeline to fully process footage from on-track cameras, ensuring seamless data collection and analysis.
• Built an injury detection system capable of identifying foreleg lameness with high accuracy.
• Horse tracklets are projected to RWC using epipolar geometry to derive key performance metrics such as speed, stride length, acceleration, and symmetry.
• Real-time alerts are sent to veterinarians via SMS within 3–5 minutes of detecting an adverse event.
• Alerts are integrated into the RH360 app, allowing vets to review cases and provide feedback.

The system successfully identified hundreds of veterinarian-confirmed foreleg injuries, demonstrating its effectiveness in injury detection. Its impact has driven strong demand from veterinarians and industry professionals seeking to enhance equine safety.