mml_face_sdk 0.1.1-dev.8

Offline, on-device 1:1 face verification and passive liveness SDK.

0.1.1-dev.8

• Restored the autoplaying demonstration GIF to the GitHub and pub.dev README while retaining links to the complete recording.

0.1.1-dev.7

• Made the Face SDK product page the primary package homepage.
• Replaced the shortened animated preview with a direct path to the complete demonstration recording.

0.1.1-dev.6

• Added a recorded enrollment, replay-rejection, and live-verification demonstration to the package documentation.
• Added optimized MP4, animated GIF, and poster media for GitHub and website presentation.

0.1.1-dev.5

• Raised the default face-similarity match threshold from 0.70 to 0.75.
• Added progressive live distance guidance with an animated arrow and proximity bar in the demo.
• Removed face-guide shapes from the recognition and liveness camera screens.

0.1.1-dev.4

• Rebuilt the demo with an embedded camera experience for registration and continuous verification.
• Added native Android NV21 camera-frame APIs for immediate stream-based enrollment, recognition, and liveness without taking photos.
• Added locally stored person names, live recognition/liveness overlays, and a dedicated recognition success screen.
• Refreshed the demonstration UI with a production-style home screen and guided face capture.

0.1.1-dev.3

• Fixed the demo enrollment flow so face actions remain disabled until activation and model installation finish.
• Added explicit enrollment state, clearer capture feedback, and guarded recognition/liveness actions until a template exists.
• Normalized Android camera JPEG orientation from EXIF metadata before face detection.

0.1.1-dev.2

• Added app-bound seven-day trials that begin on first activation.
• Added private CLI tooling to create, list, and disable trial or lifetime keys.
• Added a direct-launch Android demonstration build and public download documentation.

0.1.1-dev.1

• Renamed the package and public API to mml_face_sdk and MmlFaceSdk.
• Changed commercial licensing to lifetime, app-bound keys reusable across installations.
• Removed all biometric model weights from the publishable package.
• Added one-time encrypted private model-pack delivery during activation.
• Added app-private native model installation with tensor-shape validation.

0.1.0

• Initial pre-release Flutter API for enrollment, recognition, and passive-liveness verification.
• Native Android ML Kit/TFLite and iOS Vision/TensorFlow Lite engines.
• Offline Ed25519 seven-day trial licences and Cloudflare Worker activation backend.
• Example app, automated compatibility/licensing tests, security and release documentation.