PyTorch implementation of "HGMamba: Enhancing 3D Human Pose Estimation with a HyperGCN-Mamba Network", IJCNN2025.
The project is developed under the following environment:
- Linux
- Python 3.8.10
- PyTorch 1.12.0
- CUDA 11.6+
For installation of the project dependencies, please run:
- [Option]
pip install causal-conv1d>=1.4.0: an efficient implementation of a simple causal Conv1d layer used inside the Mamba block. pip install mamba-ssm: the core Mamba package.pip install mamba-ssm[causal-conv1d]: To install core Mamba package and causal-conv1d.
- Download the fine-tuned Stacked Hourglass detections of MotionBERT's preprocessed H3.6M data here and unzip it to 'data/motion3d'.
- Slice the motion clips by running the following python code in
data/preprocessdirectory:
For HGMamba-Base:
python h36m.py --n-frames 243
For HGMamba-Small:
python h36m.py --n-frames 81
For HGMamba-XSmall:
python h36m.py --n-frames 27
Please refer to P-STMO for dataset setup. After preprocessing, the generated .npz files (data_train_3dhp.npz and data_test_3dhp.npz) should be located at data/motion3d directory.
python main.py --config xxx.yaml
python main.py --config xxx.yaml (need to set evalue_only)
We are currently organizing and cleaning the code before making it fully available to the public.
- comming soon...
Stay tuned for updates, and feel free to ⭐ the repository for notifications!
If you find our work useful, please cite our paper:
@article{cui2025hgmamba,
title={HGMamba: Enhancing 3D Human Pose Estimation with a HyperGCN-Mamba Network},
author={Hu Cui and Tessai Hayama},
journal={arXiv preprint arXiv:2504.06638},
year={2025},
url={https://arxiv.org/abs/2504.06638}
}You can find the paper on arXiv.