samviva 1.0.0

A Flutter liveness detection library powered by Google ML Kit. Detect live users with configurable probability thresholds via active face challenges.

1.0.0

Initial release of samviva — a Flutter liveness detection library powered by Google ML Kit Face Detection.

Features

• Plug-and-play SamvivaLivenessView widget — drop into any Scaffold body; handles camera initialization, ML Kit processing, and challenge sequencing automatically
• 4 active liveness challenges powered by Google ML Kit on-device face classification:
  • SamvivaLivenessChallenge.blink — detects eye closure via eyeOpenProbability
  • SamvivaLivenessChallenge.smile — detects smile via smilingProbability
  • SamvivaLivenessChallenge.turnLeft — detects head turn via headEulerAngleY
  • SamvivaLivenessChallenge.turnRight — detects head turn via headEulerAngleY
• Configurable probability thresholds via SamvivaLivenessConfig:
  • blinkEyeClosedThreshold — lower = stricter blink detection
  • smileThreshold — higher = stricter smile detection
  • headTurnAngle — higher = must turn head further
  • challengeTimeout and totalTimeout for session control
• Per-frame probability callback via onChallengeUpdate for logging and debugging
• Optional success countdown via successCountdown — shows a 3…2…1 overlay before firing onSuccess
• Full color theming via SamvivaLivenessTheme:
  • activeColor, errorColor, overlayColor
  • stepActiveColor, stepPendingColor, connectorIdleColor
• Full text localisation via SamvivaLivenessLabels — override every user-visible string (step labels, instructions, state messages) for any language
• Full text style customization via SamvivaLivenessTextStyles:
  • titleStyle, instructionStyle, stepLabelStyle, countdownStyle
• Polished built-in UI:
  • Live camera preview with face oval cutout overlay
  • Sweeping progress arc around the oval
  • Animated step indicator with connector lines
  • Glassmorphism instruction card with animated text transitions
  • Pulse animation during active detection
  • Smooth color transitions between states
• All processing is on-device — no server, no network call
• Supports Android (NV21 image format) and iOS (BGRA8888 image format)