Case study · Face liveness & anti-spoofing SDK
- Industry
- Fintech
- Services
- Mobile Apps, AI Solutions
- Technology
- Flutter, Google ML Kit, ONNX Runtime
01 · The challenge
What stood in the way
Banks, fintechs and eKYC apps must be sure a real person is onboarding, not a printed photo, a screen or a recorded video. Cloud checks charge per verification and send face images off the device.
The check also has to be quick and clear for first-time smartphone users, in poor light, in their own language.
02 · The solution
What we built
Face Defender runs entirely on the phone and is added to Android and iOS apps in a few lines of Flutter code. A passive check takes about 5 seconds; an active check takes about 20 seconds with four short prompts. Live hints and spoken prompts in English and Bangla guide the user, and camera frames are processed in memory, never saved or uploaded by default.
Each session gets a fresh random challenge order, so a recording of an earlier session fails. A passive presentation-attack model plus depth and screen-artifact cues stop spoofing, and device checks flag emulators, root or jailbreak, virtual cameras, screen capture and debuggers.
Results are structured (verified, suspicious, failed, cancelled or error) with a risk score, confidence and reason. An NID eKYC add-on scans both sides of the national ID and matches its portrait to the live selfie, and the SDK is white-label and modular.
03 · The results
~5 s
Passive liveness check
0
Per-check fees or paid APIs
Offline
Core check works on the device
Next project
Revere Technologies
Want results like these?
Tell us about it. We reply within 24 hours with next steps and a free consultation.