smart_liveliness_detection 0.3.5

This package helps you verify that a real person is present in front of the camera, not a photo, video, or mask.

Changelog

0.3.5 April 19, 2026

New Features

3D Depth Detection (iOS TrueDepth / ARKit)

• Added DepthDetectionConfig — opt-in configuration with depthThreshold (metres), requireTrueDepth, failSessionOnSpoofing, and minFramesRequired
• Added DepthDetectionResult model with depthStdDev, depthVariance, confidence, vertexCount, and isTrueDepthAvailable
• Added DepthDetectionService — Dart platform-channel wrapper; no-ops gracefully on Android and unsupported iOS devices
• Added iOS native plugin (DepthDetectionPlugin.swift) using ARFaceTrackingConfiguration — measures Z-axis standard deviation across the ARKit face mesh (~1 220 vertices) to distinguish real 3-D faces from flat photos and screen replays
• Added SmartLivelinessDetectionPlugin.swift root registrar and ios/smart_liveliness_detection.podspec
• Depth session runs in parallel with challenges — no extra liveness phase
• Result added to antiSpoofingDetection metadata as depthSpoofDetected: bool
• LivenessController.lastDepthResult getter exposes the most recent result
• Package converted to Flutter plugin (flutter.plugin section in pubspec.yaml)

Biometric Template Generation

• Added TemplateConfig — selects BiometricAlgorithm (currently geometricRatios) and optional XOR obfuscation key
• Added BiometricTemplate model with encodedVector (base64), rawVector, algorithm, sessionId, createdAt, featureCount
• Added BiometricTemplateService — extracts ~27 normalised geometric features from ML Kit face landmarks (positions + inter-landmark ratios), serialises to Float32List → bytes → base64 with optional XOR obfuscation
• Added BiometricMatcher utility — compare() returns cosine similarity 0.0–1.0; isMatch() applies a threshold
• New LivenessConfig fields:
  • generateBiometricTemplate (default false) — opt-in flag
  • templateConfig — algorithm and obfuscation settings
  • referenceTemplate — previously enrolled template to match against
  • biometricMatchThreshold (default 0.80) — cosine similarity pass threshold
• New onBiometricTemplateGenerated callback on LivenessDetectionScreen and LivenessController
• When referenceTemplate is set, biometricMatchScore and biometricMatchPassed are injected into onLivenessCompleted metadata
• Template generation is privacy-first: no raw pixel data stored, one-way conversion

Dependency Updates

• camera → ^0.12.0+1
• camera_android → ^0.10.10+16
• google_mlkit_face_detection → ^0.13.2
• sensors_plus → ^7.0.0
• flutter_tts → ^4.2.5
• uuid → ^4.5.3
• cupertino_icons → ^1.0.9
• flutter_lints → ^6.0.0
• mockito → ^5.6.4
• build_runner → ^2.13.1
• Removed dartdoc dev dependency (conflicts with mockito ≥5.6.4 analyzer constraint; pub.dev generates docs server-side)

Other

• Fixed stale example code that had leaked into lib/main.dart and removed it
• Example app: added "3D Depth Detection", "Biometric Template — Enroll", and "Biometric Template — Verify" demo screens
• Added dartdoc comments to all Assets constants for pub.dev documentation scoring
• Moved example entry point to example/lib/main.dart (pub.dev canonical location)

---

0.3.4 April 19, 2026

New Features

Face Quality Scoring

• Added FaceQualityResult model with an overall score (0–100) and per-metric breakdown: brightness, sharpness, headPose, faceSize, eyeOpenness
• Added FaceQualityService — lightweight pixel-sampling analysis that runs on every 10th face-detected frame to avoid jank
• New onFaceQualityCheck callback on LivenessDetectionScreen — fires with each quality result so apps can surface feedback to the user
• New LivenessConfig fields:
  • enableFaceQualityScoring (default false) — opt-in flag
  • minFaceQualityScore (default 60.0) — score threshold used when blocking is enabled
  • blockChallengesOnLowQuality (default false) — when true, session stays in centering phase until quality score meets the threshold
• New LivenessMessages.lowFaceQuality — customisable message shown when score is too low
• LivenessController.lastQualityResult getter exposes the most recent FaceQualityResult

Screen Flash Anti-Spoofing

• Added ScreenFlashConfig — configurable RGB flash test that runs between face centering and the first challenge
• Added ScreenFlashResult model with passed, colorDeltas (per-color luminance response vs baseline), baselineLuminance, and confidence
• Added ScreenFlashService — internal state machine: baseline → flashRed → flashGreen → flashBlue → done
• Full-screen colored overlay rendered automatically during the test via LivenessController.activeFlashColor getter
• Camera exposure is locked (ExposureMode.locked) for the duration of the flash test to prevent AEC from cancelling the signal, then restored automatically
• Configurable warmup frames per color (warmupFramesPerColor, default 2) — skips early frames while screen and camera settle
• New LivenessConfig.screenFlash: ScreenFlashConfig? — null by default (opt-in)
• New LivenessState.screenFlashTest — inserted between centeringFace and performingChallenges
• Flash result included in antiSpoofingDetection metadata as screenFlashSpoofDetected: bool
• New LivenessMessages fields: screenFlashInstruction, screenFlashSpoofingDetected
• failSessionOnSpoofing flag controls whether a failed flash test ends the session or just sets the metadata flag

0.3.2 February 21, 2026

• Patch: Version bump and minor stability improvements

0.3.1 February 21, 2026

Bug Fixes

• Fixed a critical bug in VoiceGuidanceService where TTS would attempt to speak even when not initialized, causing crashes on some devices. Added a guard clause to prevent any TTS operations if the service is not properly initialized.
• Resolved head turning challenges not resetting the session correctly when the face was not detected, ensuring a more consistent user experience.
• Improved error handling in the camera service to prevent crashes when the camera feed is interrupted or unavailable.

0.3.0 February 21, 2026

New Features

Voice Guidance & Accessibility

• Added VoiceGuidanceConfig — fully configurable TTS settings (language, volume, speech rate, pitch, repeat interval)
• Added VoiceGuidanceService — debounced TTS wrapper built on flutter_tts that prevents audio flooding from ~30 fps camera callbacks
• Voice guidance speaks: initial instruction, face centering feedback, each challenge instruction, and completion/failure result
• Fine-grained control flags: speakPositioningFeedback, speakChallengeInstructions, speakCompletion
• Two convenience presets: VoiceGuidanceConfig.minimal() (no centering speech) and VoiceGuidanceConfig.accessibility() (slower rate, shorter repeat interval)
• Exported VoiceGuidanceConfig from the top-level package barrel
• Zero overhead when disabled — VoiceGuidanceService is only instantiated when voiceGuidance?.enabled == true

Futuristic UI Painter Styles

• Added 13 new animated canvas overlay painter styles selectable via LivenessStyle enum:
  • quantum — pulsing energy rings with particle scatter effect
  • liquidMetal — flowing chrome shimmer with metallic sheen
  • cosmos — deep-space star field with nebula gradient
  • hologram — cyan holographic scan lines and grid
  • singularity — gravitational lens distortion vortex
  • synapse — neural network node-and-edge animation
  • kinetic — motion-blur speed lines and momentum trails
  • prism — rainbow light refraction prismatic effect
  • obsidian — volcanic glass dark sheen with ember glow
  • monolith — stark geometric brutalist framing
  • chronos — clockwork gears and time-dial overlay
  • floating — soft levitating bubble particles
  • sumi — Japanese ink-wash calligraphic brushwork

Futuristic Oval Overlay (FuturisticOvalOverlay)

• Style-matched animated oval face frame with per-style border color, glow, and corner HUD brackets
• Rotating progress ring that fills as liveness challenges are completed
• Animated scan-line sweep across the face region

Liveness Style Picker (LivenessStylePicker)

• New bottom-sheet widget for switching painter styles at runtime
• Live animated mini-previews of all 13 styles rendered inside the picker

Challenge Hint Widget Enhancements

• Added ChallengeHintStyle enum with 5 visual styles: plain, glass, futuristic, minimal, neon
• Added ChallengeHintAnimation enum with 4 entrance animations: scaleIn, slideUp, bounceIn, flipIn
• Hint widget now respects both style and animation on every challenge transition

Dependencies

• Added flutter_tts: ^4.2.0

Platform Setup

• Android: Added android.intent.action.TTS_SERVICE <queries> intent to AndroidManifest.xml for Android 11+ package visibility
• iOS: Configured AVAudioSession with .playback category and .mixWithOthers option in AppDelegate.swift so TTS audio is heard even when the ring/silent switch is off

---

0.2.3 November 26, 2025

• Bug fixing and improvements:
• • Minor bug fixes

Version 0.2.2- November 25, 2025

• Improvement and new features:
• • Challenge hint animations: Display GIF/Lottie animations to guide users through challenges
• • Customizable hint positions: Choose from multiple positions (top center, bottom center, corners)
• • Per-challenge hint configuration: Configure hints individually for each challenge type
• • Optional custom animations: Users can provide their own GIF or Lottie files
• • Flexible hint display: Enable/disable hints globally or per challenge
• • Default hint animations: Built-in GIFs for blink, smile, nod, and head rotation challenges

Version 0.2.1- November 24, 2025

• Bug fixing, improvements and new features:

• This version has several improvements in anti-spoofing techniques:

• Anti-Spoofing Result Map

• Screen Glare Detection

• Motion Correlation Check

• Face Contour Analysis (Mask Detection)

• Details:

• Bug fixing: Ignoring wrong error message (errorProcessing) after session.isComplete.

• Bug fixing: Fixing glare detection method and adding option to enable/disable it.

• Improving verifyMotionCorrelation method. Now checking both X and Y axes.

• Adding params to enable/disable motion correlation detection.

• Adding mask detection feature by detection face contours (The user can choose to enable/disable this feature, as well as the number of contours detected. The user can also choose which types of challenges will be checked).

• Anti-spoofing settings: Screen reflection detection and missing facial contour detection no longer block liveness detection.

• Anti-spoofing detection is configured in the metadata under antiSpoofingDetection flags (Anti-Spoofing Result Map), without preventing successful results.

0.2.0 October 25, 2025

• Added support for new liveness challenges: "Raise Eyebrows" and "Open Mouth"
• Improved face detection accuracy with updated ML models
• Enhanced UI customization options for better theming
• Fixed minor bugs and improved overall performance

0.1.3 April 25, 2025

• Google ML Kit upgraded to version 0.11.0
• Bug fixes and improvements

0.1.1 April 24, 2025

• Bug fixes and improvements

0.1.0 April 24, 2025

• Bug fixes and improvements
• Android fix initialization fix

0.0.1-beta.5 April 23, 2025

• Bug fixes and improvements
• Android fix initialization fix

0.0.1 Initial Release (April 15, 2025)

• Initial release of the Face Liveness Detection package
• Features included:
  • Multiple liveness challenge types (blinking, smiling, head turns, nodding)
  • Random challenge sequence generation
  • Face centering guidance with visual feedback
  • Anti-spoofing measures
  • Customizable UI with theming support
  • Animated progress indicators and overlays
  • Optional image capture capability