Validation of pitchAI^TM markerless motion capture using marker-based 3D motion capture

Area of Science:

• Biomechanics
• Sports Science
• Motion Capture Technology

Background:

• 3D optical marker-based systems are the gold standard for motion capture.
• Markerless motion capture offers a potentially more accessible and efficient alternative.
• Validating new markerless systems against established methods is crucial for adoption.

Purpose of the Study:

• To compare and validate baseball pitching mechanics using a single-camera markerless motion capture solution (pitchAI™) against a 3D optical marker-based system.
• To assess the accuracy of pitchAI™ in quantifying joint angles and spatiotemporal parameters during pitching.
• To determine the reliability of pitchAI™ as a markerless alternative for pitching analysis.

Main Methods:

• Ten healthy pitchers performed maximum effort fastballs while captured by both marker-based and pitchAI™ systems.
• Time-series kinematic data were compared using R-squared (r²) and root mean square error (RMSE).
• Discrete kinematic and spatiotemporal parameters were analyzed using descriptive statistics, Bland-Altman analysis, and correlation coefficients.

Main Results:

• Time-series joint angles showed high correlation (r² 0.69–0.98) and acceptable RMSE values.
• Bias for discrete parameters ranged from -11.31 to 12.01.
• RMSE for arm speed, stride length, and ball visible metrics were within acceptable ranges for practical application.

Conclusions:

• The pitchAI™ markerless motion capture system demonstrates strong agreement with 3D optical marker-based systems for quantifying pitching kinematics.
• pitchAI™ is recommended as a viable markerless alternative for analyzing baseball pitching mechanics.
• Establishing a database of pitchAI™ ranges is suggested for future inter-system comparisons.