Intel Corp (NASDAQ:INTC) has announced an expansion of its RealSense 3D cameras family with the launch of an on-devices facial recognition system using its RealSense depth-sensing tech. The system is meant to conduct facial authentication of consumer-facing solutions such as ATMs, kiosks, and point-of-sale systems.
RealSense facial recognition tech
The company said that the RealSense Id camera system will combine a specialized neural network and active depth sensor to perform facial recognition. Intel introduced RealSense 3D tech cameras in 2014 as Kinect-Style cameras for touch-free interaction. But it seems RealSense ID is the company’s attempt to reposition the RealSense 3D camera business towards on-device facial authentication which could potential put it in the controversial crossfire. The repositioning of the tech sees the company employ the core RealSense tech and package it in a way it can be used in secure access control and retail settings.
However, this could court controversy to the company considering facial recognition systems have greatly been criticized. Opponents of facial recognition tech argue that the technology could invade privacy and potentially presents Orwellian-style scope. The company is committed to getting ahead of such criticism and has promised that its system will be p[privacy driven and has been purpose-built to protect users.
RealSense tech works once prompted by a pre-registered user
Intel explained that they have created the RealSense ID tech with ant-spoofing technology which will block any attempted false entry like the use of videos, photography, or masks. They tout it to have a one-in-1-million false acceptance rate with all facial images processed locally and user data encrypted. Equally the solution can only be actives via user awareness and cannot authenticate unless a pre-registers user prompts it. Intel says that as will all technology they are committed to ensuring the ethical application of technology and human rights protection.
The company says that it has taken measures to address the possible bias of the tech by training it with diverse samples of faces. Intel said they collected extensive data from all ethnicities from Europe, Asia, and Middle East Africa.