Automated Step Detection in Inertial Measurement Unit Data From Turkeys

Area of Science:

• Animal Science
• Biotechnology
• Engineering

Background:

• Locomotion is a key indicator of turkey welfare and health.
• Current breeding values for locomotion rely on subjective scoring.
• Sensor technologies offer objective evaluation of turkey gait.

Purpose of the Study:

• To compare three step detection methods using inertial measurement units (IMUs) for turkey gait analysis.
• To identify the most accurate method for precise start and end point detection of turkey steps.
• To establish a foundation for objective turkey locomotion evaluation.

Main Methods:

• Turkeys equipped with IMUs on their upper legs walked through a corridor.
• Manual annotation of turkey steps was performed.
• Change point detection, local extrema approach, and gradient boosting machine were evaluated for step detection and precision.

Main Results:

• All three methods successfully detected steps, but the local extrema approach had more false detections.
• Gradient boosting machine demonstrated the highest precision in identifying step start and end points.
• Gradient boosting machine achieved a precision of 0.81 and recall of 0.84 at a 0.2s tolerance, significantly outperforming other methods.

Conclusions:

• Gradient boosting machine is the most accurate method for signal segmentation in turkey gait analysis using IMUs.
• This method requires an annotated training dataset for optimal performance.
• Objective gait analysis can enhance turkey welfare and health assessments.