Researchers with the University of Western Australia have developed a new 3D facial recognition model that is now available on an open basis to other researchers.
Called FR3DNet, the model was trained on 3.1 million 3D scans of over 100,000 individuals. Now, given images depicting a range of positions, expressions, and so on, the system can match them to its set of ‘known’ individuals. As the model’s developer, Dr. Syed Zulqarnain Gilani, explained to UWA University News, “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.”
The model could find some popularity in the mobile sector as 3D facial recognition becomes increasingly popular. Apple made a huge splash with its launch of the Face ID system in the iPhone X last autumn, and other major smartphone makers are starting to follow suit. Having an open platform like FR3DNet could help some of these players to significantly reduce the costs associated with development of such technologies.
FR3DNet was detailed in paper published in the Computer Vision and Pattern Recognition journal, credited to Dr. Gilani and associate researcher Ajmal Mian.
Source: UWA University News
–
July 9, 2018 – by Alex Perala
Follow Us