face_detection_camera 0.3.1

Smart Flutter camera with face detection, look-straight constraints, auto-capture, and customizable overlay.

Changelog

0.3.1

New Features

• Per-challenge image capture: Each LivenessChallengeResult now includes a capturedImage field, captured at the moment the challenge completes or fails.
• Configurable challengeDelay: Delay between challenges is now customizable via LivenessCameraWidget.challengeDelay (default: 500ms).

Improvements

• Safer image processing: _processImageIsolate now deletes the file after successful decode instead of before, preventing permanent image loss on decode failure.
• LivenessResult.capturedImage is now set from the last challenge's image (no extra capture call).

Bug Fixes

• Liveness capturedImage always null: LivenessController._completeFlow() now captures an image via captureCallback before creating LivenessResult, populated on both pass and fail.
• autoStart = false unusable: Added a "Start" button UI when autoStart is false, and exposed startLiveness() method on _LivenessCameraWidgetState.
• Crash after dispose (async _completeFlow): Added _isDisposed guard throughout LivenessController to prevent accessing disposed ValueNotifiers after async operations.
• Crash from Future.delayed after pop: _successChallenge now guards the 500ms delay with _isDisposed and state checks.
• _holdTimer passes challenge without face: Added onFaceLost() method; LivenessCameraWidget calls it when face becomes null, cancelling the hold timer.
• Missing app lifecycle handling: Added WidgetsBindingObserver to LivenessCameraWidget to pause camera & cancel timers on background.
• Anti-spoofing too strict: Replaced instant-fail on tracking ID change with tolerant maxTrackingIdChanges (default: 3).
• Capture state collision: FaceCameraController.capture() now uses a Completer so concurrent calls wait for the ongoing capture instead of returning silently.
• Wrong default threshold for turn/nod: LivenessChallengeConfig.threshold now defaults to 20° for turn/nod challenges (was 0.5, which meant any tiny movement would pass).

0.3.0

New Features

• Liveness Detection: Integrated a flexible banking-app style liveness flow with customizable challenges (blink, smile, turn left/right, nod up/down).
• LivenessCameraWidget: New specialized widget for handling the entire liveness flow with built-in UI and animations.
• Bilingual Documentation: Added comprehensive API comments in both English and Vietnamese.

Architectural Improvements

• Modular Refactoring (SRP): Decomposed the monolithic controller and widget into specialized services:
  • CameraService: Manages camera lifecycle, flash, and zoom.
  • FaceDetectionService: Handles ML Kit integration and face data processing.
  • StabilityTracker: Manages face stability and capture countdown.
  • LivenessController: Orchestrates challenge sequencing and flow.
• Sub-widget Extraction: Moved complex UI components into modular sub-widgets for better maintainability.

Bug Fixes & Optimizations

• App Lifecycle Handling: Fixed camera issues when moving the app to background/foreground.
• Memory Management: Ensured proper disposal of all controllers and ValueNotifiers to prevent leaks.
• Double File Deletion: Resolved a redundant file delete operation during the capture flow.
• Improved Metrics: Refined FaceQuality calculations with a more stable brightness estimation.
• Enhanced Customization: Exposed all internal thresholds, colors, icons, and builders as parameters.

0.2.0

New Features

• Multiple Face Detection: Added detectMultipleFaces, maxFaces, and onMultipleFacesDetected callback
• Face Quality Score: Added FaceQuality class with brightness, sharpness, and pose metrics
• Manual Capture Button: Added showCaptureButton and captureButtonBuilder for custom UI
• Preview Mode: Added showPreview, onPreviewConfirm, and onPreviewRetry callbacks
• Zoom Control: Added pinch-to-zoom with enableZoom, minZoom, maxZoom, and zoomLevel notifier
• Capture Sound: Added enableCaptureSound option

Optimizations

• Error Recovery: Auto-retry camera initialization with exponential backoff (maxRetryAttempts)
• Memory: Proper disposal of all ValueNotifiers and resources

0.1.0

• Initial public release with basic face detection and auto-capture.