Synchrony Vision: Capturing Body Motion Synchrony Through Phase Difference Using the Kinect

Abstract

This study introduces a Kinect-based system for real-time detection of body motion synchrony, addressing limitations of traditional methods such as high costs, intrusiveness, subjectivity, and lack of real-time analysis. Using a phase difference detection algorithm, the system analyzes synchrony offering a comprehensive framework for quantifying interpersonal coordination. The system captures acceleration data from up to six individuals, calculates phase differences, and provides immediate feedback. The original algorithm was modified to address challenges such as gravitational acceleration, inverted axes, and peak detection thresholds, improving the accuracy of Kinect data analysis. Experimental results demonstrate the system’s overall detection accuracy of 89.2% under controlled conditions. Additionally, a comparison with an accelerometer-based method revealed a strong correlation (r = 0.73, p = 0.002), indicating alignment in detecting synchrony. While the Kinect-based system offers advantages in scalability and usability, it exhibits limitations in detecting high-speed movements. To enhance its accuracy, potential improvements include algorithmic refinements, hardware upgrades, and AI-driven models to adaptively refine motion detection.