Feasibility and Accuracy of an RTMPose-Based Markerless Motion Capture System for Single-Player Tasks in 3x3...

Area of Science:

• Sports Science
• Biomechanics
• Performance Analysis

Background:

• Markerless motion capture (MMC) is a non-invasive technique for monitoring external load in sports.
• Its application and accuracy in 3x3 basketball contexts are not yet established.
• Wearable devices are often restricted in basketball, highlighting the need for alternative monitoring methods.

Purpose of the Study:

• To evaluate the viability and measurement precision of a multi-camera RTMPose-based MMC system for single-player 3x3 basketball tasks.
• To validate the system's accuracy in capturing key movement metrics relevant to basketball performance.
• To determine if MMC can reliably track player movements in an outdoor basketball setting.

Main Methods:

• Eight cameras (60 fps) recorded ten collegiate athletes performing basketball-specific drills (sprints, T-tests, jumps) on a half-court.
• 3D coordinates of hip and ankle keypoints were reconstructed using Direct Linear Transformation (DLT).
• Horizontal displacement and average speed were derived and validated against tape-measure and time-motion analysis.

Main Results:

• The MMC system showed high accuracy with coefficients of variation (CVs) under 5% and mean bias below 3.5%.
• Speed estimates demonstrated excellent consistency with time-motion analysis (ICC = 0.97-1.00).
• Bland-Altman analysis confirmed minimal bias and no proportional error.

Conclusions:

• The RTMPose-based MMC system is a consistent and accurate non-invasive method for collecting movement data in 3x3 basketball.
• This technology can be effectively used for single-player performance monitoring in outdoor basketball environments.
• Future research should investigate system performance during live play with multiple athletes and compare it to laboratory-grade systems.