Evaluating a Multi-Camera Markerless System for Capturing Basketball-Specific Movements: An Exploration Using 25 Hz...

Area of Science:

• Biomechanics
• Motion Analysis
• Sports Technology

Background:

• Markerless motion capture (MMC) offers a non-invasive approach to motion analysis.
• The validity of MMC at 25 Hz, common in broadcast and surveillance, is not well-established for sports applications.

Purpose of the Study:

• To evaluate the performance of a 25 Hz multi-camera MMC system using consumer cameras for basketball movements.
• To assess the accuracy of kinematic data derived from MMC at a standard broadcast frame rate.

Main Methods:

• Six consumer-grade MMC cameras (DJI Action 5 Pro, 25 fps) and a 10-camera Vicon system synchronously recorded athletes performing basketball tasks.
• Kinematic data were processed using an RTMDet-RTMPose pipeline and filtered at 6 Hz.
• Waveform validity was assessed using Pearson's correlation coefficient (r) and root mean square error (RMSE).

Main Results:

• Excellent agreement was found for joint displacement magnitudes (r = 0.916-0.994; nRMSE = 0.54-1.32%).
• Agreement decreased for velocity (r = 0.583-0.867) and acceleration (r = 0.232-0.677), indicating sensitivity to the low sampling rate.
• The 25 Hz MMC system showed limitations for high-precision impact analysis but acceptable accuracy for macroscopic tracking.

Conclusions:

• A 25 Hz MMC workflow provides acceptable accuracy for tracking joint displacement and quantifying external loads in resource-limited settings.
• The low sampling rate significantly impacts the accuracy of derivative kinematic variables (velocity, acceleration).
• Prioritizing temporal synchronization is crucial for improving the reliability of derivative variables in future MMC systems.