Detecting artificial facial images

Detecting artificial facial images 

We are seeking partners for:

-Technology licensing or assignement

-Several algorithms can be
provided

For Contact:

Technology Transfer:
INNOVECTIS GmbH
Altenhöferallee 5
60438 Frankfurt am Main, Germany
Phone: +49 69 25 61 632-0
matthias.goetz@innovectis.de

---

Technology

This technology enables Morphing Attack Detection (MAD) for identifying artificial facial images created by blending the faces of at least two different subjects.

It includes differential detection methods that compare a questioned image with a trusted probe image, using both facial landmark analysis and deep face representations.

Applications

The solution is designed for identity verification workflows in which facial images are used for official or automated recognition purposes, especially in electronic travel documents and biometric control systems. It is particularly relevant for photo-credential issuance agencies, companies, and organizations that need to detect morphed images capable of deceiving automated or manual face recognition systems:

-Border control

-Travel document verification

-Identity verification

 -Biometric recognition systems

 Technology stage

-The underlying algorithms have been successfully developed and tested, and the project status states that proof of concept has been conducted very successfully.

-The technology is described as best in class according to the NIST FRVT MORPH performance test, with several related studies already published, including DeepFace-TIFS-2020 and LandmarkMAD-ICISP.

Patents

–The intellectual property includes granted EP3642756B1 and US11238271B2 with priority date June 20, 2017.