3D Human Kinematics and Dynamics From a Single Smartphone Video
OpenCap-Monocular estimates 3D human movement kinematics and musculoskeletal kinetics from a single smartphone video, combining pose estimation, camera optimization, and biomechanical modeling.
- Single-camera processing
- 3D pose estimation (WHAM)
- Camera and pose optimization
- OpenSim integration
- Activity classification (walking, sit-to-stand, squats, …)
Requires Ubuntu 20.04/22.04, Python 3.9, NVIDIA driver ≥ 520.
git clone https://github.com/utahmobl/opencap-monocular.git --recursive
cd opencap-monocularSee installation/INSTALL_SLIM.md for step-by-step instructions.
- Video preprocessing & rotation correction
- WHAM 3D pose estimation
- Camera extrinsics & pose optimization
- OpenSim IK and export
- Visualization (
mono.jsonfor OpenCap Visualizer)
mono.json— OpenCap viewer*.trc,*.mot— OpenSim formats*_scaled.osim— Scaled model
WHAM · ViTPose · SLAHMR · OpenCap · OpenSim · VideoLLaMA3
This project is licensed under the PolyForm Noncommercial License 1.0.0 — non-commercial use only.
Commercial use requires a separate agreement. Contact the authors for inquiries.
This project depends on third-party components with their own licenses. Some impose additional restrictions:
| Component | License | Commercial Use |
|---|---|---|
| WHAM | MIT | Allowed |
| ViTPose | Apache 2.0 | Allowed |
| DPVO | MIT | Allowed |
| SMPL / SMPL-X body models | Custom (MPI) | Not allowed |
The SMPL body models required by this pipeline are licensed by the Max Planck Institute for non-commercial scientific research only. Users must agree to the SMPL license before downloading the model files. For commercial SMPL licensing, contact ps-license@tue.mpg.de.