Liveness
Liveness overview
Prove a real, live person is in front of the camera.
Liveness verification answers a single question: is there a real, live human in front of the camera right now, or is the stream a photo, a video replay, a 3D mask, or a synthetic deepfake?
iLive's liveness product combines anti-spoof, deepfake detection, face-consistency checks, and motion analysis into a single managed flow. The output is a verdict you can act on, plus an ICAO-compliant photo of the verified person you can store as your reference image.
Verdicts
Every session ends with one of three verdicts:
| Verdict | Confidence | What it means | Suggested action |
|---|---|---|---|
pass | ≥ 0.70 | Confident live person. | Proceed with onboarding / login. |
retry | 0.45 – 0.70 | Inconclusive — lighting, motion, or coverage issue. | Ask the user to try again once. |
fail | < 0.45 | High likelihood of spoof or deepfake. | Reject and flag for review. |
The exact confidence cutoffs above are the platform defaults; the verdict string is always authoritative for your decisioning.
Two modes
- Active — the user completes a few interactive challenges like a blink or a head turn. Highest security, slightly more friction.
- Passive — the user holds still for about three seconds. Lowest friction, still runs the full anti-spoof stack. Recommended for onboarding flows.
Where to go next
- Managed sessions — the recommended server-to-server flow for KYC and onboarding.
- ICAO photo — retrieving and using the
passport-quality photo produced after a
pass. - Attestation — the integrity pattern we use today and our roadmap for signed verdicts.
- KYC onboarding guide — end-to-end recipe combining liveness with face compare and dedupe.