The rage of the pandemic doesn't seem to slow down, yet the world economy's nose-diving state is demanding the reopening of operations. As the world is looking forward to resuming daily activities, the focus is on how to keep every interaction in the slightest. Even in the pre-COVID world, automation, virtualizations, and touchless remote solutions were gaining traction for the convenience they provide, but now, they are the need of the hour. Just like how collaborative virtual meeting platforms have become an essential amenity, so would the touchless solutions in the post lockdown phase.
Adopting a touchless access control solution is one such measure; every workplace and office is now looking for. So far, most workplace entry systems like biometric fingerprint scanners, PIN pads, keycards, etc. were touch-intensive. If the same system continues to be used, there would be the need to use disposable wipes after every touch. A touchless access control solution would enable the workplaces to do away with this exhaustive process. In the following sections, some of the available touchless access control technologies and solutions are described. The proposed solutions, alongside preventing direct contact, ensure much lesser scopes of impersonation and forgery when compared with the existing solutions.
UWB based Tags
UWB based tags that are available as wristbands or ID cards can give the exact locations of the people carrying them. Such tags, when integrated with an access control unit, can provide an automatic touchless access control solution.
Figure 1 – UWB based Touchless Access Control System – PathPartner
Why choose UWB over other wireless technologies?
- High precision ranging
- Less chance of interference with other wireless technologies
- Able to penetrate different materials
- Better resilience to multipath
AI-based Facial Recognition
Facial recognition is the term used for the process of recognizing a human face with the use of technology. With the help of a camera or a scanner (2d or 3d) and advanced algorithms, unique facial features are mapped and detected. The various facial features considered are the distance between the two eyes, width, and shape of the forehead, the distance between the forehead and the chin, width of the lips, nose, and eyebrows.
The facial recognition market is expected to grow to $7.7 billion in 2022 from $4 billion in 2017.
From the perspective of technology, all these parameters that represent the unique characteristics are nothing but data points or nodal points. These data can be stored as well as accessed whenever needed. The unique facial geometry gives rise to a 'facial signature.' Each facial signature is compared to the database of known faces and that's how the availability of a match is discovered.
Figure 3 – Facial Recognition System – PathPartner
This is, essentially, the same way that iris scanners and fingerprint scanners work, but uniquely programmed to work with facial recognition. To compensate for unusual camera angles or poor image quality, the facial recognition algorithm can detect these angles and make the adjustments necessary to the faceprint.
From the technology point of view, it’s not very different from how iris scanners or fingerprint scanners work. It’s just programmed to work with facial features. The robustness of an FR system is determined by how accurately the algorithms can cancel out the impact of poor image quality and unusual angles.
The increased popularity of facial recognition is also attributed to its ease of use. All a person has to do is stand in front of the camera or scanner and leave it to the system to identify the unique facial signature and decide whether to unlock the assigned door. Time taken for such decisions are usually quick, say within milliseconds. Such security systems are trouble-free yet more secure. Needless to say, people wearing gloves need to go through the hassle of having to remove them.
But, will it work on masked faces?
Experts state that facial recognition technology may not work accurately for obscured faces as less nodal points would be present to make a comparison.
Google's facial recognition system for Pixel 4 was built to identify a person's face, even in the absence or presence of a beard or sunglasses. But later, when tested with face masks, the Face Unlock didn’t work. Also, Apple's Face ID, which prides itself as one of the early adopters of facial recognition systems for smartphones, faced similar complaints about the iPhones launched in 2017. During the wildfire season of 2018, when the Californians had to keep their faces covered by masks were among the complainers.
The issue lies at the heart of how the majority of the algorithms for recognizing faces are designed. In order to come up with a robust, error-free identification, as many as possible aspects of a face are considered. Thus, not just eyes and forehead, everything from the nose, mouth, shape of the face, chin are considered.
As per a claim from the researchers of Wuhan University, 95% accuracy was achieved in recognizing masked faces.
Well, the good news is, our algorithm works on masked faces!
In the post COVID world, even though the technologies enabling touchless access control were gaining immense popularity, they were never considered to be replacing the existing access control systems completely. But, the demand for minimum possible contacts is changing the complete landscape of access control systems. With the current growth rate in machine learning and computer vision algorithms, we might see some radical advancement in the field of facial recognition technology, which might accurately identify masked faces.
- How Radar technology can make UV LED sterilizers safer and more efficient?
- How sensors can ensure safety in public places?
- How deploying an RTLS in your workplace can be a lifesaver?
- Sensors – The most vital element of the Internet of Things
- Your ultimate guide to choosing the right UWB methodology
- Advantages of using Ultra-Wideband (UWB) technology for Indoor Positioning