Intel launches RealSense ID , its own Face ID with which they seek to improve all those problems that many of the facial recognition systems that exist in the market currently suffer. For this reason, it is committed to the use of depth sensors and a single database with which to feed its neural networks.
Privacy-focused facial recognition
Facial recognition technology is much more convenient than other biometric security systems in most situations. For example, taking away from the fact that we now all have to wear a mask, unlocking a mobile phone just by lifting it up and facing you is much more agile than depending on placing your finger in a specific place so that it recognizes the fingerprint.
The big problem with facial recognition is that few are the manufacturers that have managed to implement a really safe and efficient system in terms of possible physical changes that we all suffer over time. Like changes in the haircut, use or not of glasses or other accessories, grow a beard, etc.
Well, Intel wants or intends to end all these problems and also add extra value by focusing development on user privacy and much more precise recognition, without the problems already seen by similar proposals. For this, it not only launches a new facial recognition system called RealSense ID , a version of Apple‘s Face ID that is used in the iPhone 12 for example, it also improves it with the use of technologies that will be key.
To begin with, RealSense ID is a facial recognition system that makes use of depth sensors . Here we already have an important improvement since it prevents the system from being deceived with a simple photograph. Something that we have already verified can happen on mobile phones. By not being able to detect the depth of the image, a photo of the user would serve the correct distance to access any system that uses facial recognition based on the use of a simple camera sensor.
On the other hand, RealSense ID would also make use of neural networks and a dedicated chip that would process all that information securely on the device itself. This chip, in addition to avoiding dependence on a network connection, would also avoid possible manipulations by not allowing your data to be accessed from the outside.
Of course, the most important thing seems to be how the neural networks that Intel would use here have been fed with data. And it is that we have all witnessed the problems that somewhat erratic recognition systems can create if the identification is not as precise as it would be desired. For example, there have been cases where the Amazon cameras that US police officers have been using tested positive for people who were not really who the system was telling them.
To suppress these false positives, Intel has created a unique user base, analyzing thousands and thousands of faces of different ethnic groups from different continents. In this way, from Asia to Europe, Africa or the Middle East, the idea is to be sure to cover any possible variant that may exist in order to have a much more reliable system.
Safe recognition for professional and personal use
RealSense ID could be a great improvement for the use of a multitude of computers and other applications in both professional and personal environments. Because the idea of Intel is not to build something that we would only see in ATMs, customs controls or the like. This technology would also have applications for everyday use such as smart door locks, etc.
So that it works and that it works the way Intel promises is something that could benefit us all. Because other options could be used, but taking into account everything we already have, it is true that continuing to advance in improvements in everything related to facial recognition is as necessary as finding a precise way to create a voice profile capable of identifying uniquely to each user. Although the latter will be much more complicated in the short term.
Finally, RealSense ID will be priced at $ 99 and will be released during the first quarter of 2021. So it is likely that we will see its integration in various devices sooner than we might initially think. And its size, smaller than a credit card although somewhat thicker, will allow its integration more easily.
