Home

Awesome

PhysMamba

Codes of paper "PhysMamba: Efficient Remote Physiological Measurement with SlowFast Temporal Difference Mamba" Arxiv Accepted by Chinese Conference on Biometric Recognition (CCBR 2024 Oral)

<img src="figures/Pipeline.png">

Set up

Code is based on rPPG-Toolbox

conda create -n PhysMamba python=3.8
pip install torch==2.0.0 torchvision==0.15.1 torchaudio==2.0.1 --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt
cd causal-conv1d
python setup.py install
cd mamba
python setup.py install

Datasets

Train

Cross-dataset

  1. Modify the configuration file ./configs/train_configs/UBFC-rPPG_UBFC-rPPG_PURE_PHYSMAMBA
  2. Run python main.py --config_file ./configs/train_configs/UBFC-rPPG_UBFC-rPPG_PURE_PHYSMAMBA.yaml
  1. Modify the configuration file ./configs/train_configs/PURE_PURE_UBFC-rPPG_PHYSMAMBA_BASIC.yaml
  2. Run python main.py --config_file ./configs/train_configs/PURE_PURE_UBFC-rPPG_PHYSMAMBA_BASIC.yaml

Intro-dataset on UBFC-rPPG

  1. Modify the configuration file ./configs/train_configs/UBFC_INTRO_PHYSMAMBA.yaml
  2. Run python main.py --config_file ./configs/train_configs/UBFC_INTRO_PHYSMAMBA.yaml

Citation

@inproceedings{luo2024physmamba,
  title={PhysMamba: Efficient Remote Physiological Measurement with SlowFast Temporal Difference Mamba},
  author={Chaoqi Luo and Yiping Xie and Zitong Yu},
  booktitle={Chinese Conference on Biometric Recognition (CCBR)},
  year={2024}
}