mwpose3d is a 3D human pose estimation toolbox built on PyTorch, specifically designed for mmWave radar point cloud data. It provides a modular and extensible framework for developing, training, and evaluating 3D human pose estimation models. The toolbox is built on top of OpenMMLab's mmengine framework. By leveraging mmengine, mwpose3d offers a flexible configuration system and a standardized training and evaluation pipeline suitable for both research and applied scenarios.
mwpose3d can be installed as a Python package following the installation guide.
The following documents describe the end-to-end workflow for dataset preparation, model training, and evaluation:
- Public dataset preparation
- Custom dataset preparation
- Gather tracking records
- Training and evaluating models
mwpose3d also provides several application-level tools: