Face SDK Features: Liveness
Is it a real human being looking at the camera? Not a mannequin or a 3D image? The Liveness verification module is created to check whether the biometric information source accessing the camera is a physically present live person.
In addition to enhancing security measures against spoofing attacks, deepfakes, and fraud, the liveness verification module offers an extra capability. By leveraging advanced AI algorithms, it can also provide an estimated age of the person with an accuracy of approximately +/-3 years.
Liveness verification can be conveniently performed from a device with a camera, where the person simply needs to follow the animated instructions. During the verification process, the system records a video session and analyzes it to determine both the person's liveliness and estimate their age.
Warning
The liveness verification can be performed only using a camera with a resolution of HD 1024x720 or higher.
Tag
When starting, you can specify tag which all attempts to read liveness will be bound to. If you don't specify it, the application will generate it automatically.
The calling application returns tag and transactionId. One session may include many transactions, and from a transaction ID it is clear which session it belongs to.
You can get the metadata (device model, screen data, frame size, app ID and version, OS version, platform, SDK version, etc.), person's selfie, and video of the liveness detection session. They will be saved on the backend at the following path:
faceapi-session/year={year}/month={month}/day={day}/hour={hour}/minute={minute}/{tag}/transactionId.
To access liveness transaction artifacts, use the GET /api/v2/liveness?transactionId={transactionId} OpenAPI request: See OpenAPI documentation.