aws-amplify / aws-amplify/amplify-ui-android
Face Liveness: how to set face-distance / "Move closer" threshold for a walk-up kiosk (standing distance)?
- Dominant language
- Kotlin
- Stars
- 25
- Forks
- 18
- Avg merge
- 7d 3h
- Merged PRs (30d)
- 1
Description
### Which UI component?
Face Liveness (`FaceLivenessDetector`, Android)
### The use case
We're deploying `FaceLivenessDetector` on a **self-service walk-up kiosk** — a 22" Android tablet (Elo i-Series 5), app locked to portrait. Guests **walk up and stand at a normal kiosk distance** (roughly 40–80 cm from the screen), rather than holding a phone at arm's length.
### The problem
During the FaceMovement challenge the component repeatedly prompts **"Move closer"** and only proceeds once the user brings their face very close to the camera — essentially nose-to-screen. That's fine for a handheld phone selfie, but it's impractical for an unattended kiosk where people naturally stand back.
Our working theory is that the built-in tablet camera is wide-FOV, so a face at standing distance occupies only a small share of the frame and never reaches the oval-fill / face-distance threshold the challenge requires.
### Questions
1. **Is the face-distance / oval-fill threshold configurable** (how much of the frame the face must fill before the challenge advances), so the check can complete at a normal standing distance without requiring the user to move very close?
2. If it isn't configurable, what are the **recommended best practices / camera requirements** for running Face Liveness on an **unattended kiosk at standing distance** — e.g. camera field-of-view, resolution, or a recommended external camera setup?
3. Is a **passive** Face Liveness mode (single frame, no "move closer" challenge) available or on the roadmap for kiosk / stand-off use cases?
### Environment
- `com.amplifyframework.ui:liveness` 1.10.0
- Android on a landscape-native 22" tablet (Elo i-Series 5), activity locked to portrait
- Amazon Rekognition Face Liveness, eu-west-1
Any guidance on kiosk / standing-distance deployments would be really appreciated. Thanks!
Contributor guide
Research direction
Start with the FaceLivenessDetector API and the Amazon Rekognition Face Liveness documentation, focusing on face-distance configuration, kiosk camera requirements, and passive-mode support. Done means establishing whether these options exist in version 1.10.0 or are on the roadmap, and documenting the supported guidance for the described kiosk setup.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- android, aws, kotlin
- Domain
- mobile, security
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100