Here we summarize the most important aspect regarding the 2 main approaches for liveness detection.
Definition
Active liveness: challenge-response detection, where the user is asked to perform some actions:
Head movements (Truora)
Blinking
Following a dot on screen
Smiling
Speaking random numbers
Moving the camera toward one’s face
The performed analysis aims to detect the motions.
Passive liveness: requires no action from the user. The liveness detection occurs when the user takes a selfie.
It's based in various techniques:
Analyzing selfie
Capture a short video (motion detection)
Project lights on the subject
Depth sensing (hardware-assisted)
Texture analysis
The performed analysis aims to detect the spoof artifact.
User Experience (UX)
Active liveness: challenge-response action is required, takes longer and might be confusing. Higher drop off rates are reported.
Passive liveness: requires no action from the user, cleaner experience and resulting in lower drop off rates.
Speed
Active liveness: depends on the challenge made to the user, however processing time tends to be longer.
Passive liveness: tends to be much faster than active solutions.
Image analysis
Active liveness: requires multiple frames to detect user’s action.
Passive liveness: Depends on the approach/methodology used.
Security
Active liveness: provide fraudsters with information on how the liveness check is done (i.e. info on how to defeat the system).
Passive liveness: fraudsters don’t know there is a liveness check taking place, is harder for them to know how to defeat the system.
Main restrictions
Passive liveness: high quality images are needed.