flutter_bio_auth 0.1.0

On-device face detection, tracking, active liveness challenges, and an injectable 1:1 face-recognition / anti-spoof pipeline for Flutter. Detection/tracking/liveness work with zero bundled AI models ( [...]

Changelog

All notable changes to this project will be documented in this file.

0.1.0 Initial Release

Dart API

• Added the core FaceBiometric API.

• Added FaceBiometricController for managing authentication flows.

• Added FaceBiometricConfig with configuration validation.

• Added comprehensive result, error, and authentication-state types.

• Added pure-Dart reference implementations for:

  • FaceMatcher
  • ChallengeSequencer

• Added unit tests covering the core Dart components.

Android

• Added ML Kit-based:

  • Face detection
  • Face tracking
  • Face landmarks
  • Active liveness detection

• Added CameraX-based camera processing pipeline.

• Added injectable TensorFlow Lite interfaces for:

  • FaceEmbeddingEngine
  • AntiSpoofEngine

• Added fail-closed Unavailable* implementations when required AI models are unavailable.

• Added Android Keystore-backed encryption for biometric templates.

• Added AES-GCM encrypted local template storage using SharedPreferences.

• Added explicit authentication gates through AuthenticationEngine.

iOS

• Added Vision-based:

  • Face detection
  • Face tracking
  • Face landmarks
  • Active liveness detection

• Added AVFoundation-based camera processing pipeline.

• Added injectable Core ML interfaces for:

  • FaceEmbeddingEngine
  • AntiSpoofEngine

• Added fail-closed Unavailable* implementations when required AI models are unavailable.

• Added iOS Keychain-backed encryption for biometric templates.

• Added AES-GCM encrypted local template storage using UserDefaults.

• Added explicit authentication gates through AuthenticationEngine.

Security & Privacy

• Designed biometric processing to run locally on the device.

• Added encrypted biometric template storage.

• Added platform-secure key management using:

  • Android Keystore
  • iOS Keychain

• Added fail-closed behavior when required recognition or anti-spoofing components are unavailable.

• No raw face images are required for biometric template storage.

Known Limitations

The following items remain incomplete in this initial release:

• compare() still-image detection entry point is currently stubbed.
• iOS lookUp / lookDown pitch support is currently incomplete.
• Recognition and anti-spoofing require compatible AI models to be provided or integrated.
• The implementation still requires validation against real Flutter, Android, and iOS production toolchains and physical devices.

See the Status / Limitations section in the README for additional details.