Researchers develop model to scale 3D facial recognition
Researchers from The University of Western Australia (UWA) have developed a model for scaling 3D facial recognition which they say could transform the biometrics industry.
A team from the UWA Department of Computer Science and Software Engineering led by Dr. Syed Zulqarnain Gilani created a model called FR3DNet, which it says is the first of its kind, by analysing 3.1 million 3D scans of more than 100,000 individuals. The model was trained to identify a large dataset of known faces, and then match a test face to one of them.
“With off-the-shelf 3D cameras becoming cheap and affordable, the future for pure 3D face recognition does not seem far away,” Dr Gilani said. “Our research shows that recognition performance on 3D scans is better and more robust. Your 3D scan could be in any pose, wearing glasses or a face mask, and laughing or just smiling and the deep model can recognise you in an instant. We hope that this research will help improve security on devices that use facial recognition to grant access to networks and systems.”
The results of the research have been published in the Journal of Computer Vision and Pattern Recognition. The FR3DNet model is currently available for research purposes.
While most 3D facial recognition applications so far are based on small scale applications, like the solutions unveiled recently by Megvii and Vivo at MWCS18, SensibleVision released its 3DSafe high-volume authentication system in April.