e841018 / remark

Privacy-preserving Fiducial Marker System via Single-pixel Imaging

ReMark

This repository provides the source code used in ReMark: Privacy-preserving Fiducial Marker System via Single-pixel Imaging accepted in MobiCom 2023. Please cite our paper if it helps your research.

Requirements

• Python 3.6 or newer

Installation

1. Clone this repository.
2. Install Python packages: <python> -m pip install -r requirements.txt
3. Install a PyTorch version according to your computing platform.

Usage

• Identify marker ID from observations:

  cd src
  <python> demo.py

  • prints details and shows reconstructed images.

• Synthesize datasets:

  cd src
  <python> synthesize.py

  • generates dataset/recon_detection/clear_detection_sample=100000.pkl

    and dataset/recon_identification/clear_identification_sample=100000.pkl

• Train reconstruction NNs:

  cd src
  <python> train_recon_detection.py 0

  • generates model/recon/model=[dcan_M=72]__train=[clear_detection_sample=100000]__rep=0/model.pt

  cd src
  <python> train_recon_identification.py 0

  • generates model/recon/model=[dcan_M=256]__train=[clear_identification_sample=100000]__rep=0/model.pt

• Train alignment NNs:

  cd src
  <python> synthesize_and_reconstruct.py

  • generates dataset/align/dcan256_snr=15db_sample=100000.pkl

  cd src
  <python> train_align.py 0

  • generates model/align/model=[cnm]__train=[dcan256_snr=15db_sample=100000]__rep=0/model.pt

Notes on SPI hardware

This repository does not include the code that drives the SPI hardware in the paper. It is highly recommended to replace the outdated DMD and USRP with modern hardware, in which case the old code will not be suitable. Some observations collected by the outdated hardware are available in the data directory. The formats are documented in demo.py.

Citation

@inproceedings{10.1145/3570361.3613289,
    author = {Yu, Tzu-Hsu and Tsai, Hsin-Mu},
    title = {ReMark: Privacy-Preserving Fiducial Marker System via Single-Pixel Imaging},
    year = {2023},
    isbn = {9781450399906},
    publisher = {Association for Computing Machinery},
    address = {New York, NY, USA},
    url = {https://doi.org/10.1145/3570361.3613289},
    doi = {10.1145/3570361.3613289},
    booktitle = {Proceedings of the 29th Annual International Conference on Mobile Computing and Networking},
    articleno = {74},
    numpages = {15},
    keywords = {soft-decision decoding, single-pixel imaging, fiducial marker, retroreflective imaging, singularity-free embedding},
    location = {Madrid, Spain},
    series = {ACM MobiCom '23}
}