Synthetic Vascular Spoofing Sample Database (aka Synthentic Vascular Proesentation Attack Instrument (PAI) Database)

Andreas Vorderleitner, Jutta Hämmerle-Uhl, and Andreas Uhl: ''Finger Vein Spoof GANs: Can we Supersede the Production of Presentation Attack Artefacts?'' Proceedings of the 22th International Workshop on Digital Forensics and Watermarking (IWDW'23), Springer LNCS 14511, pp. 109-124, 2023.

This page provides a download link for synthetic data that have been generated to produce the results given in the paper above. In addition to the synthetic data, we also provide insight how the data have been generated (which model, which parameters, organisation of training data etc.) to enable reproducability.

These data have also been used in a series of follow-up publications, for which we also define which parts of the data have been used to produce the results in those further papers.

All those papers together with the datasets used are detailed in the file README_FIRST.txt, which also contains a list of all further documents / files contained in the repository (including an explanation of the file structure of the synthetic data and the sample file naming conventions used).

Contact:

• Andreas Uhl (Maintainer)

This dataset contains synthetic finger vein and hand vein sample data which have been generated using the spoof (aka presentation attack instrument (PAI)) data contained in the original PLUS, IDIAP, SCUT, and PROTECT datasets in order to study their applicability in presentation attacks (i.e. vulnerability analysis and PAD training). Note that the original genuine and spoof samples are not contained but need to be retrieved from the original providers of the datasets.