Intel has introduced RealSense ID, a facial-authentication system the company unveiled during CES week. It combines an active depth sensor with a neural network for on-device face recognition and anti-spoofing, and Intel is pitching it at device makers for smart locks, access control, point-of-sale systems and ATMs.
How it works
RealSense ID uses a depth camera rather than relying only on a 2D selfie image. The depth sensor captures three-dimensional information about the shape of a face, and that 3D data is fed to a neural network which performs the recognition task and checks that the input really is a live person.
Because the system includes both depth sensing and neural-network processing, it’s designed to spot common spoofing attempts such as photos, videos or masks. The depth map helps distinguish a flat image from a real face, while the on-device neural model looks for signs of life and other anti-spoofing cues.
Intel emphasises that the biometric processing happens on the device itself. Templates and recognition logic are stored and run locally, rather than being sent to a cloud service for matching, which is intended to reduce privacy concerns and exposure to data breaches.
The company also says the system can adapt to gradual appearance changes — for example facial hair, glasses or ageing — by updating the stored templates over time so users aren’t constantly asked to re-enrol.
Why this matters for devices and businesses
Face authentication is already familiar from phones, but Intel’s approach is aimed at business and embedded use-cases where privacy, robustness and integration matter. For smart locks and office access control, on-device processing means an entry system can make matches without transmitting biometric data across a network. For payment terminals and ATMs, the anti-spoofing features are particularly important because they add a layer of fraud protection beyond a simple 2D camera.
Because Intel is offering a packaged solution — depth sensor plus recognition software — it’s a product for device makers rather than something consumers will install themselves. The target customers are manufacturers of doors, kiosks, unattended payment devices and other pieces of hardware that need reliable, private biometric identity checks.
What to watch next: compatibility and integration. Vendors will want to know how RealSense ID fits into existing access-control ecosystems, what development tools Intel provides, and how straightforward it is to deploy across many devices. There’s also the usual trade-off space around speed, power use and cost for an embedded depth camera plus local processing.
If you’re running hardware that needs a robust, privacy-aware way of recognising people, Intel’s pitch is that RealSense ID brings depth-aware, AI-driven facial authentication into devices that can’t — or shouldn’t — rely on cloud-based biometric services.
This article has been restored to the What's New On The Net archive as part of the site's relaunch.