18 Oct The evolution of facial recognition software
Facial recognition software has vastly improved in the last decade yet there’s still a perception that it’s an inferior form of security and can cause more problems than it solves. ATG Digital’s Ariel Flax is keen to bust those myths. He discusses how far the technology has come against some previous obstacles.
Windows
Today’s facial recognition cameras can even operate when faced with a glass barrier like a window. Flax adds: “On top of that, the software improves its accuracy and speed as it learns your face over time. It can positively identify from just a part of the face.”
This development is especially useful when it comes to tackling issues of rain or glare. The time taken to read a face falls from seconds to mere milliseconds.
Masks
With mask wearing so prevalent in light of Covid, you may think facial recognition software would struggle to operate when faces are obscured by masks. But think again. A partial scan is enough in order for a positive ID to be obtained. Not only that but Flax advises systems can be used to enforce mask wearing compliance. Facial recognition systems can be programmed to only grant access to those who are wearing a mask correctly.
Glare, reflection and low light
Again, the fact that partial recognition capabilities are possible helps in situations like these. Flax also advises: “It’s best to have a roof overhead wherever the reader is mounted to benefit the visitor and the equipment.” By mounting devices undercover reflection can be blocked.
Rain
Rain – especially rain on a window – can distort an image which historically has meant poor results from facial recognition tech. Today, the technology has become so advanced that they software can recognise a face with only a fragment of an image. Flax does caution however, that in order to do so the technology will need to learn over time.
Internet connectivity
No one wants their entire access control system to rely on connectivity – but facial recognition software doesn’t need to. Readers can be mounted on an arm which is connected to a permanent ethernet cable.
You can also go one step further with a system that works offline. Flax explains that this solution works in a similar way to LRP technology that can store large amounts of database information locally on the device.
Remote visitor loader
Biometric access control solutions usually need someone to physically capture the data and load it – not so with facial recognition. Although it is a time-consuming task it is possible to load all existing residents, tenants and staff. For long stay visitors or contractors facial data can be loaded via digital photographs.
“This functionality can be easily integrated into Protech apps, enabling contractors and guests to consent and upload their image from their phones. No fuss on the day of arrival” he explains.