FakeIDet3-DB: A Database of Digital ID Manipulations for Forensic Model Training
A team of researchers has launched FakeIDet3-DB, the inaugural extensive database focused on digital alterations of authentic, government-issued identity documents. This collection features traditional manipulations, including copy-move, alongside Generative AI techniques like face-swapping and inpainting, all improved with sophisticated image refinement to minimize visual artifacts. Its purpose is to connect the gap between synthetic templates and actual IDs for developing resilient image forensic models. To adhere to data protection laws such as GDPR, a patch-based framework is utilized. The findings are documented in arXiv:2607.26641.
Key facts
- FakeIDet3-DB is the first comprehensive database of digital manipulations on real, government-issued IDs.
- It includes classical manipulations (e.g., copy-move) and Generative AI-driven manipulations (e.g., face-swapping, inpainting).
- Manipulations are enhanced with advanced image refinement procedures to suppress visual artifacts.
- The database adopts a patch-based framework to comply with GDPR and other data protection regulations.
- The research is published on arXiv with ID 2607.26641.
Entities
—