Oops, that didn't work. Please try again later.

Liveness Detection: The Missing Piece for Infinitely Scalable Facial Authentication

Back

Aug. 19, 2019, 4:26 p.m.

When a person gains access to a secure building, sensitive data, or vast sums of corporate finance via facial authentication, how do you know they are who they say they are - for sure?

the realface photo

Masks from The Real Face Japan. 

Yes, Chooch AI has a neural network model and learns the 512 biometric features of every face it is trained to learn. Yes, when an image of a face is sent via API to Chooch AI, we have a higher than 99% accuracy in facial authentication. We know whether that face has been learned by Chooch or not, but what happens if the person trying to gain access has a very, very good mask? 

Only liveness detection ensures that facial authentication is being performed on a living face, and up until now, liveness detection has been imperfect. In a recent evaluation for a bank ranked in the top 100 worldwide with an expected initial user base of six million people, Chooch AI didn’t just perform near flawlessly for facial authentication. The bigger breakthrough is that Chooch AI also detected if the face was not on a human head or was on a video. Even when exposed to a near perfect mask or a high resolution digital display of a video feed of a face, Chooch detected the absence of liveness, again more than 99% of the time. Close to 100%, actually.

To achieve this, our team layered multiple machine learning models, combining aspects of image recognition and object detection to create a validation process that outputs a simple true or a false signal, this is a living face or not. If a face isn’t blinking, for example, it’s not live. Go ahead, build a mask that blinks and Chooch AI will detect that, too. 

closeup eye

Mask from The Real Face Japan. 

Spoofproof facial authentication is now infinitely scalable with liveness detection and is camera agnostic.  You can use any type of camera for facial authentication now. When Chooch AI is trained to recognize a face, with as little as one image, it creates an undecryptable piece of math known as a hash. That unique number represent the 512 biometric features which the Chooch AI facial recognition algorithm calculates, but the identity of the person cannot be reverse engineered. This keeps identities safe.

Chooch can make safe calculations for millions of people in real time. This means any top 100 global bank can now use facial authentication to drastically reduce fraud. Account holders can now be assured that only their face, along with PIN numbers, SMS messages, and so on, will radically increase their security by unspoofable mulitfactors.

Keep in mind that this application is not for facial recognition in state surveillance purposes, and also, it’s not only to ensure that authenticated people can send money from one bank to another. Facial authentication can now ensure an individual who is signing a document electronically is actually the one signing it, not an assistant. It can ensure the bitcoin transaction is by the true account holder, and not by a “friend”, or a hacker who hijacked a SIM card. It’s to make sure that a person opening a door to a secure facility actually has a security clearance, and not just a mask from Mission Impossible which Chooch AI could, of course, detect.

Chooch AI is performing these operations in the cloud on a network of GPUs. There are now no hardware costs to doing facial authentication with liveness detection, and it works agnostically with any camera. Simply install an API, train the system with your faces, and you can use our service to authenticate. Chooch AI is also deployable on premise to an existing secure GPU network on a private cloud, which means that liveness detection can become part of your global financial transaction infrastructure, even for credit card authorizations which occur millions of times every second. 

Get in touch when your institution would like a demo.

a man 1

Mask from The Real Face Japan.