20252_1
Prototypischer Ansatz zur 3D-Pose-Schätzung mit etablierter Computer-Vision-Methode
This project implements markerless 3D human pose estimation by spatially fusing
monocular 2D keypoints detected via MediaPipe across a multi-camera system. By
leveraging wide-angle lenses and a high-precision ChArUco calibration, the system
enables robust, real-time spatial reconstruction in full-room environments.
(Marie Bohnert / Prof. Dr. Schwanecke, WS 25/26, bestes Poster)