On this page
Why does liveness detection matter?
Liveness detection is a security check that confirms a real, live person is at the camera. Face recognition cannot do this on its own. A matcher answers whether two faces are the same face, and a photograph of the right person passes that test perfectly.
So the attack is cheap. A picture taken from a public profile, held up to a laptop camera, defeats recognition alone. Liveness is the control that decides whether the sensor is looking at a person or at a copy of one.
How does liveness detection work?
Every method looks for evidence that the thing in front of the lens has the depth, texture, and behavior that a flat or manufactured surface does not. The two families differ mainly in whether the user is asked to do anything.
Passive methods have largely won on user experience. Active methods still have a place where a slower, more deliberate check is acceptable.
Passive liveness reads the capture
The check runs on the frame itself, using texture, reflection, and the way light behaves on real skin. The user does nothing extra, so nothing is added to sign-in time.
Active liveness asks for a movement
A blink, a smile, or a head turn proves a response in real time. It costs a few seconds and it makes a replayed video harder to use.
Testing separates the claims
Any vendor can assert liveness. What distinguishes an implementation is an independent laboratory result against a defined attack set, not the technique named on the datasheet.
How does liveness detection relate to presentation attack detection?
The two terms describe one goal in two vocabularies. Presentation attack detection, or PAD, is the formal name. Liveness detection is what people say in a product meeting.
The formal side is worth knowing when you read a datasheet. ISO/IEC 30107-1 holds the taxonomy, and ISO/IEC 30107-3 specifies how PAD is tested and reported. A vendor that cites 30107-3 is pointing at a test method, which is a stronger claim than a description of a technique.
How does SenseCrypt use liveness detection?
SenseCrypt signs a user in by face, so liveness is not a bolt-on risk signal. It runs as part of the same on-device ceremony as the match, and a capture that fails it does not become a sign-in.
The SenseCrypt liveness holds iBeta ISO 30107-3 certification at Level 1 and Level 2. Capture is ordinary 2D RGB, which is why it works on the phone a user already carries.
It runs where the capture happens
Liveness and matching both run on the user's own device. Nothing leaves that device unless the ceremony passes.
Certified at Level 1 and Level 2
An accredited laboratory tested it against low-cost artifacts such as printed photos and screen replays, and against custom-made artifacts such as a fitted 3D mask.
No special hardware
The check works from a standard 2D RGB camera. There is no depth sensor, so enrollment is not gated on a particular phone tier.
Frequently asked questions
What is liveness detection?
Liveness detection is a security check that confirms a real, live person is at the camera. It separates a live face from a photo, a video, or a mask.
Does liveness detection stop a photo attack?
Yes. Liveness detection checks that a face is real and present. It rejects a printed photo or a screen video.
Is liveness detection the same as presentation attack detection?
The two terms share one goal. Presentation attack detection (PAD) is the formal term in the ISO/IEC 30107-3 standard. Liveness detection is the common name.
Does SenseCrypt use certified liveness detection?
Yes. The SenseCrypt liveness detection holds iBeta ISO 30107-3 certification. It covers Level 1 and Level 2 presentation-attack detection.
Related