PyTorch implementation of "HGMamba: Enhancing 3D Human Pose Estimation with a HyperGCN-Mamba Network", IJCNN2025.

🛠️ Environment

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.

📂 Dataset

Human3.6M

Preprocessing

1. Download the fine-tuned Stacked Hourglass detections of MotionBERT's preprocessed H3.6M data here and unzip it to 'data/motion3d'.
2. Slice the motion clips by running the following python code in data/preprocess directory:

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

MPI-INF-3DHP

Preprocessing

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.

🏋️‍♂️ Training

python main.py --config xxx.yaml

🧪 Testing

python main.py --config xxx.yaml (need to set evalue_only)

🔧 Code Release in Progress

We are currently organizing and cleaning the code before making it fully available to the public.

📅 weight:

• comming soon...

Stay tuned for updates, and feel free to ⭐ the repository for notifications!

📄 Citation

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.