A Multi-Planar Computer Vision Framework for Low-Cost 3D Motion Capture Reconstruction and Gait Analysis
Description
This dataset and computational framework introduce a high-performance, video-based 3D motion capture (MoCap) and gait analysis system designed as a low-cost alternative to marker-based optoelectronic hardware (e.g., Vicon, Qualisys). By processing standard 2D video sequences through markerless computer vision pipelines, the system extracts, filters, and reconstructs multi-planar human skeletal motion across three distinct anatomical planes: Coronal, Sagittal, and Transverse.Methodological Architecture & Technical Specifications Multi-Planar Pose Reconstruction: Utilizes the MediaPipe Pose tracking topology to extract 33 spatial body coordinates from single-camera 2D digital video streams. The framework projects these trajectories into a unified coordinate system to simultaneously visualize human locomotion across orthogonal anatomical views.Signal Filtering and Noise Mitigation: To eliminate spatial tracking jitter and edge noise inherent in markerless vision models, an Exponential Moving Average (EMA) smoothing algorithm is implemented. This produces mathematically stable trajectories required for precise joint angular velocity and acceleration calculus. Pseudo-Calibration Matrix: Implements a robust baseline calibration protocol that standardizes the subject's coordinate metrics (height and depth fields) using dynamic anthropometric averages, ensuring consistent depth perception (Z-axis scaling) relative to pixel dimensions. Visual Differentiation: The codebase generates dynamic graphics featuring perspective-based edge rendering and distinct segmentation for critical anatomical endpoints (such as the head and terminal foot landmarks) to isolate points of primary biomechanical interest. Potential Reuse and Scientific Applications:This research repository serves sports scientists, physical therapists, clinical biomechanists, and computer vision engineers. The codebase and trajectory data can be utilized to evaluate clinical gait abnormalities, validate markerless tracking frameworks against laboratory gold standards, or develop field-based athletic movement diagnostics without expensive laboratory constraints. Core Academic Foundations Reflected:Winter, D. A. (2009). Biomechanics and Motor Control of Human Movement.Whittle, M. W. (2014). Gait Analysis: An Introduction.Lugade, V. et al. (2011).
Files
Institutions
- Sabaragamuwa University of Sri LankaSabaragamuwa Province, Ratnapura