PXL Vision and Idiap develop deepfake detector

PXL Vision, the Swiss premium provider of digital identity verification, and the Idiap Research Institute, one of Europe's leading institutions for artificial intelligence and biometrics, have successfully completed their joint research project to improve the detection of deepfakes. The collaboration, which began in spring 2024, focused on the identification of synthetic faces in digital identity verification.

These AI-generated images of supposedly real people are becoming increasingly widespread and pose an enormous challenge for identity verification. This is because they are now almost indistinguishable from real images. The research project was supported by the Swiss Innovation Agency Innosuisse.

Three types of manipulation recognizable

The so-called deepfake detector is now integrated as a new key technology in the PXL Ident software solution from PXL Vision and is available to all customers. The PXL Deepfake Detector complements existing fraud prevention technologies and is designed to identify three types of tampering: Face swapping, which is the replacement of one person's face with another person's; face impersonation, which is the manipulation of facial expressions or movements of a real person; and fully synthetic faces, which are AI-generated images of people that do not exist in reality.

The technology has been available to all customers since the beginning of the year. It is based on advanced deep learning architectures and has been trained on a diverse, extensive dataset of real and manipulated facial images. The deepfake detector represents an important milestone in the company's mission to ensure trust, authenticity and compliance in digital identity verification.

The challenge of identity documents

Sébastien Marcel, Head of the Biometric Security and Privacy Group and member of the interim management team at Idiap, says: «As part of the project, we have developed advanced AI-based technologies to detect and defend against deepfakes and other forms of identity fraud. This achievement reflects Idiap's long-standing expertise in biometrics, AI security and trusted digital identity. Our collaboration with PXL Vision has been both productive and inspiring. We look forward to continuing to push the boundaries of robust, privacy-compliant identity verification in the future.»

The biggest challenge for Sébastien Marcel and his team was to use their expertise in the field of biometrics to solve a new, demanding problem: ID documents. «We had limited experience with ID documents and were able to build on PXL's in-depth expertise in this area,» Marcel continues. «The result achieved exceeded our initial expectations.»

«Working with the Idiap team has been extremely rewarding,» says Nevena Shamoska, CTO of PXL Vision. «Thanks to their world-class expertise in biometrics and generative AI, we were able to significantly accelerate our progress in integrating state-of-the-art deepfake detection capabilities into our platform.»

Further investments planned

PXL Vision will continue to invest in research on biometric security, AI forensics and deepfake resilience to effectively protect customers in the financial, telecommunications and government sectors from next-generation digital identity attacks. Idiap itself will scientifically process the findings from the research project and use them as content for white papers and specialist conferences.

More information: www.pxl-vision.com

This article originally appeared on m-q.ch - https://www.m-q.ch/de/pxl-vision-und-idiap-entwickeln-deepfake-detektor/