Cattle Body Size Measurement Based on DUOS-PointNet+.

Area of Science:

• Agricultural Engineering
• Computer Vision
• Animal Science

Background:

• Non-contact livestock body size measurement using point cloud data is often complex and error-prone.
• Accurate body measurements are crucial for livestock management, breeding, and health monitoring.

Purpose of the Study:

• To develop an efficient and accurate cattle body measuring system using point cloud data.
• To introduce a novel algorithm, dynamic unbalanced octree grouping (DUOS), for improved segmentation and measurement.

Main Methods:

• A cattle body measuring system integrating the DUOS algorithm (based on PointNet++) and an efficient measurement method.
• The system segments cattle into seven parts (body, legs) and identifies key measurement points.
• Techniques include density measurement, point cloud slicing, contour extraction, and point cloud repair.

Main Results:

• The DUOS algorithm enhanced segmentation accuracy by 0.53% and mean intersection by 1.21% compared to other models.
• Relative errors for manual vs. experimental measurements were: withers height (1.18%), hip height (1.34%), body length (2.52%), thoracic circumference (2.12%), abdominal circumference (2.26%), and cannon circumference (2.78%).

Conclusions:

• The proposed system demonstrates effective segmentation of cattle bodies from point cloud data.
• The system is suitable for accurate, non-contact cattle body size measurement, offering improved precision over existing methods.