Azure AI Vision SDK 0.18.0-beta.1 (preview)

[Breaking Changes] The following changes are included in this release.

For the Mobile (iOS and Android) solution:

1. Introducing new "Passive-Active" mode support.
2. Introducing a new API surface that bundles the UI to simplify integration.

For the Web solution:

1. Updated API surface to improve the developer experience.
2. Algorithm improvements to ensure “PassiveActive” phase is more user friendly.
3. Runtime improvements to reduce the package size and fixes for runtime errors.

Please refer to the following documentation for more information on how to integrate the newly released SDKs into your applications:

• iOS: sample readme file
• Android: sample-readme file
• Web: readme file

Azure AI Vision Web SDK 0.17.2-beta.3

Refresh of the Vision Face Client Web SDKs with the following changes

Refresh of the Vision Face Client Web SDKs with the following changes

• Remove the need for setting CORS headers while serving faceanalyzer-assets.
• Reduced memory usage and binary file size for the assets.

Fixed bugs:

• Session goes to "Unexpected" as soon as it starts sometimes.
• Session goes to "FaceTrackingFailed" as soon as it starts sometimes.

Azure AI Vision Web SDK 0.17.2-beta.2 (preview)

Refresh of the Vision Face Client Web SDKs with the following changes

• Remove the need for setting CORS headers while serving faceanalyzer-assets.
• Reduced memory usage and binary file size for the assets.

Known bugs:

• Session goes to "Unexpected" as soon as it starts sometimes.
• Session goes to "FaceTrackingFailed" as soon as it starts sometimes.

Note: This release is not suitable to production yet due to the known issues. We will be releasing a fix soon by the first week of October.

Azure AI Vision SDK 0.17.2-beta.1 (preview)

Refresh of the Vision Face Client SDKs with the following changes:

For Web SDK:

• Fix for a large camera preview size issue repro'ing on various Firefox browsers.
• Reduced memory usage to fix some OOM issues.

For Mobile SDKs:

• Support for latest SDK requirements set by the Google and Apple app stores.

Mobile Face Liveness: Azure AI Vision SDK 0.17.1-beta.1 (preview)

Refresh of the Azure AI Vision Face Liveness detection for Mobile (iOS and Android) with the following changes:

• New feature to "Retry" Face Liveness session with same token. A token can be used to perform 2 retries.
• Improved UX to add "Retry" button.

Sample and Documentation Changes:

• Updated kotlin and swift samples for release 0.17.1-beta.1
• Changelog: 0.17.0-beta.1...0.17.1-beta.1

Azure AI Vision SDK 0.17.0-beta.1 (preview)

Introducing a new Liveness Web SDK in Public Preview.

The new SDK release now allows developers to utilize face liveness checks in web-browsers on both mobile and desktop devices for identity-verification scenarios. The liveness solution is designed with both security and developer experience in mind, handling all the internal complexities of performing and evaluating face liveness checks accurately. This enables developers to easily plug and play the liveness check into their applications, which significantly speeds up the development process.

By default, the liveness solution uses a hybrid "passive or active" technique. It is designed to require active motion only in poor lighting conditions, thereby enhancing the speed and efficiency of liveness checks in optimal lighting. The active motion technique extends the lighting operational range, allowing the solution to function effectively across various lighting conditions. Both techniques are highly optimized to minimize end-user friction during the liveness check operation.

The liveness solution complies with ISO/IEC 30107-3 PAD (Presentation Attack Detection) standards, as validated by iBeta level 1 and level 2 conformance testing. It effectively defends against various spoofing methods, including paper printouts, 2D/3D masks, and spoof presentations on phones and laptops. Liveness detection is an active area of research, with ongoing enhancements to counter increasingly sophisticated spoofing attacks. Continuous improvements will be rolled out to both client and service components as the solution is strengthened against new types of attacks.

Highlights of the Web SDK release:

• The new SDK combines both the algorithms and the UI implementations into a web-component. This allows developers to easily integrate the liveness check into existing web-applications with only a few lines of code. Popular JavaScript framework such React, Angular and Vue.js has been tested to work well with the Web SDK.
• The SDK also allows for customizations in terms of colors and fonts to allow developers to customize the look and feel of the solution.
• The web solution is tested across a large range of browsers as summarized below that specifies the minimum OS version number supported in each device and browser combination. This is an area that will have continuous improvements over time.
  

Release notes for the Mobile SDK also included in this release:

• Improved Apple App Store readiness.

References:

• Release blog with demo video: Blog
• To get started: Face Liveness Tutorial
• NextJS Sample: Readme.md

Mobile Face Liveness: Azure AI Vision SDK 0.16.3-beta.3 (preview)

Refresh of the Azure AI Vision Face Liveness detection for Mobile (iOS and Android) with the following changes:

• Improved Google Play Store and Apple App Store readiness.
• Added support for Face API Service private preview versions.
• Dropped support for iOS 9-11. The minimum supported version is iOS 12.
• Improved multithreading stability.

API Changes:

• iOS:
  • FaceAnalyzedResult.serviceResponseStatusCode: type changed from NSNumber* to NSInteger
  • FaceRecognitionFailureReason: 2 new enum values:
    • .faceMouthRegionNotVisible
    • .faceWithMaskDetected
• Android
  • RecognitionFailureReason: 2 new enum values
    • FACE_MOUTH_REGION_NOT_VISIBLE
    • FACE_WITH_MASK_DETECTED

Sample and Documentation Changes:

• Updated Kotlin and Swift samples for release 0.16.3-beta.3.
• Fixed Kotlin sample that may display verification image with wrong rotation in certain devices.
• Changelog: 0.16.2-beta.2...0.16.3-beta.3

Mobile Face Liveness: Azure AI Vision SDK 0.16.2-beta.2 (preview)

Refresh of the Azure AI Vision Face Liveness detection for Mobile (iOS and Android) with the following changes:

• Improved end-user positioning messages to dynamically ensure that the end-user comes closer in brighter environments to improve the success-rates.
• Improved compliance with Google App Store and Apple App Store rules for permissions and privacy-related APIs.
• Added support for confidential result mode where results are only available through querying Face API service.
• Improved Detect Liveness With Verify mode to enable verifying hashes of VerifyImage between client app and Face API service.
• Improved stability especially during teardown.

API Changes:

• iOS:
  • VisionServiceOptions.advanced is no longer nullable.
  • VisionSource.init(data:) new API to preserve VerifyImage hash.
• Android
  • VisionSource.fromByteBuffer(ByteBuffer) new API to preserve VerifyImage hash.

Sample and Documentation Changes:

• Updated kotlin and swift samples for release 0.16.2-beta.2
• Improved GET_FACE_ARTIFACTS_ACCESS.md
• Changelog: 0.16.1-beta.1...0.16.2-beta.2

Mobile Face Liveness: Azure AI Vision SDK 0.16.1-beta.1 (preview)

Refresh of the Azure AI Vision Face Liveness detection for Mobile (iOS and Android) with the following changes:

• Multiple stability fixes.
• Support multiple new features available in the Azure AI Vision Face service in the client SDK.

Mobile Face Liveness: Azure AI Vision SDK 0.16.0-beta.1 (preview)

Initial public preview release of Azure AI Vision Face Liveness detection for Mobile. See Microsoft documentation for an overview of Azure AI Vision Face Liveness Detection.

This SDK supports two feature variants:

• Liveness with Verification
• Liveness

Liveness detection aims to verify that the system engages with a physically present, living individual during the verification process. This is achieved by differentiating between a real (live) and fake (spoof) representation which may include photographs, videos, masks, or other means to mimic a real person.

The new Face liveness detection solution is a combination of mobile SDK and Azure AI service. It is conformant to ISO/IEC 30107-3 PAD (Presentation Attack Detection) standards as validated by iBeta level 1 and level 2 conformance testing. It successfully defends against a plethora of spoof types ranging from paper printouts, 2D/3D masks, and spoof presentations on phones and laptops. Liveness detection is an active area of research, with continuous improvements being made to counteract increasingly sophisticated spoofing attacks over time, and continuous improvement will be rolled out to the client and the service components as the overall solution gets hardened against new types of attacks over time.

While blocking spoof attacks is the primary focus of the liveness solution, paramount importance is also given to allowing real users to successfully pass the liveness check with low friction. Additionally, the liveness solution complies with the comprehensive responsible AI and data privacy standards to ensure fair usage across demographics around the world through extensive fairness testing. For more information, please visit: Empowering responsible AI practices | Microsoft AI.

Please see the readme documents listed below for instructions on how to build and run each sample.

iOS Sample Description: README.md

Android Sample Description: README.md