Job Description

Description

of the subject

The existing public CDP datasets [3,4] contain the CDP images and its counterfeits created via some estimation attacks. These datasets were created in the laboratory conditions and do not reflect the real-world case of CDP use. Additionally, the estimation methods used do not consider the recent advances in the field of generative neural networks and probabilistic latent diffusion models [2]. During this internship we will work on construction of more challenging CDP dataset and estimation attacks to reverse the distortions added by Printing-and-Digitalization (PD) process to create a more challenging CDP counterfeits.

The main tasks of this internship are:

  • To create a challenging real-world CDP dataset using the set of pre-defined printers and scanners, as well as smartphones.
  • To create a counterfeited CDPs using SOTA methods [3,4].
  • To propose an estimation attack based on probabilistic diffusion mo...

    • Apply for this Position

    Ready to join Université Lumière Lyon 2? Click the button below to submit your application.

    Submit Application