GaFaR

Description

This dataset includes the reconstructed face images from facial templates extracted from face images of the MOBIO dataset using template inversion methods proposed in the following paper, including GaFaR, GaFaR+GS, and GaFaR+CO. In addition, as described in the paper, the reconstructed face images were used for practical presentation attacks. To this end, each of the reconstructed face images was printed on a typical paper or shown by a digital tablet (Apple iPad Pro) and presented in front of a camera. The captured image by the camera can then be used as input to the face recognition system. The cameras of three different mobile devices were used for capturing images in this dataset, including Apple iPhone 12, Xiaomi Redmi 9 A, and Samsung Galaxy S9. In addition to the proposed face reconstruction methods, reconstructed face images by two other methods from literature are used for presentation attacks using replay from iPad and captured by iPhone 12 camera.

You can find more information about the dataset (including source code of reconstructing face images) on the project page: https://www.idiap.ch/paper/gafar/

Reference

If you're using this dataset, please cite the following publication

  @article{tpami2023ti3d,
    author    = {Hatef Otroshi Shahreza and S{\'e}bastien Marcel},
    title     = {Comprehensive Vulnerability Evaluation of Face Recognition Systems to Template Inversion Attacks Via 3D Face Reconstruction},
    journal   = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
    year      = {2023},
    volume={45},
    number={12},
    pages={14248-14265},
    doi={10.1109/TPAMI.2023.3312123}
  }