AI-Powered Livestock Measurement

Non-contact cattle measurement on an iPad, combining pose estimation with LiDAR depth to compute real-world body dimensions. Domain: Agriculture.

The challenge

USDA researchers need non-contact livestock body measurements at scale, but manual measurement is slow, invasive, and inconsistent.

The approach

iPad LiDAR + YOLOv11 pose estimation fused with depth-to-pixel conversion to compute real-world distances between anatomical keypoints, with CVAT and Unity Perception synthetic data powering the training pipeline.

About this project

This demonstration shows an advanced computer vision system that automatically measures cattle dimensions using pose estimation and depth sensing technology on iPad. The system calculates real-world measurements between anatomical points by combining pose detection with depth-to-pixel conversion algorithms.

Technology and methods

  • YOLOv11
  • iPad LiDAR
  • Unity3D
  • OpenCV
  • CVAT

Related work