This research explored the design of a non-invasive markerless vision-based system to accurately estimate a climber's Center of Mass (CoM) on an indoor bouldering wall. By utilizing consumer-grade hardware and computer vision, the system provides 3D data-driven feedback previously limited to expensive laboratory environments.
Results
Final Grade: 8.5 / A. Completing and undertaking this thesis was incredibly rewarding as it combined my technical background with my passion for climbing. Navigating the challenges of 3D motion capture in such a high-occlusion environment with fast motion and limited resources. The project is available on GitHub for anyone to download and use.
Tools
Python, OpenCap, YOLOv8, ViTPose, 3D Motion Capture
