ML Kit Release Notes

Latest Versions

iOS API Latest Versions

Artifact name	Version	Last Update
GoogleMLKit	3.0.0	04/26/2022
MLKitBarcodeScanning	2.0.0	04/26/2022
MLKitCommon	6.0.0	04/26/2022
MLKitDigitalInkRecognition	2.0.0	04/26/2022
MLKitEntityExtraction	1.0.0-beta7	04/26/2022
MLKitFaceDetection	2.0.0	04/26/2022
MLKitImageLabeling	2.0.0	04/26/2022
MLKitImageLabelingCommon	4.0.0	04/26/2022
MLKitImageLabelingCustom	2.0.0	04/26/2022
MLKitLanguageID	3.0.0	04/26/2022
MLKitLinkFirebase	2.0.0	04/26/2022
MLKitObjectDetection	2.0.0	04/26/2022
MLKitObjectDetectionCommon	4.0.0	04/26/2022
MLKitObjectDetectionCustom	2.0.0	04/26/2022
MLKitPoseDetection	1.0.0-beta8	04/26/2022
MLKitPoseDetectionAccurate	1.0.0-beta8	04/26/2022
MLKitPoseDetectionCommon	1.0.0-beta8	04/26/2022
MLKitSegmentationCommon	1.0.0-beta6	04/26/2022
MLKitSegmentationSelfie	1.0.0-beta6	04/26/2022
MLKitSmartReply	2.0.0	04/26/2022
MLKitTextRecognition	1.3.0	06/29/2021
MLKitTextRecognition (v2)	1.4.0-beta4	04/26/2022
MLKitTextRecognitionCommon	1.0.0-beta4	04/26/2022
MLKitTextRecognitionChinese	1.0.0-beta4	04/26/2022
MLKitTextRecognitionDevanagari	1.0.0-beta4	04/26/2022
MLKitTextRecognitionJapanese	1.0.0-beta4	04/26/2022
MLKitTextRecognitionKorean	1.0.0-beta4	04/26/2022
MLKitTranslate	2.0.0	04/26/2022
MLKitVision	4.0.0	04/26/2022

Android API Latest Versions

Group ID	Artifact name	Version	Last Update
com.google.android.gms	play-services-mlkit-barcode-scanning	18.0.0	01/25/2022
com.google.android.gms	play-services-mlkit-face-detection	17.0.1	02/22/2022
com.google.android.gms	play-services-mlkit-image-labeling	16.0.7	01/25/2022
com.google.android.gms	play-services-mlkit-image-labeling-custom	16.0.0-beta3	01/25/2022
com.google.android.gms	play-services-mlkit-language-id	17.0.0-beta1	01/25/2022
com.google.android.gms	play-services-mlkit-text-recognition	18.0.0	01/25/2022
com.google.android.gms	play-services-code-scanner	16.0.0-beta1	05/10/2022
com.google.mlkit	barcode-scanning	17.0.2	01/25/2022
com.google.mlkit	camera	16.0.0-beta3	01/25/2022
com.google.mlkit	digital-ink-recognition	18.0.0	01/25/2022
com.google.mlkit	entity-extraction	16.0.0-beta4	01/25/2022
com.google.mlkit	face-detection	16.1.5	02/22/2022
com.google.mlkit	image-labeling	17.0.7	01/25/2022
com.google.mlkit	image-labeling-custom	17.0.1	01/25/2022
com.google.mlkit	language-id	17.0.3	01/25/2022
com.google.mlkit	linkfirebase	17.0.0	01/25/2022
com.google.mlkit	object-detection	17.0.0	01/25/2022
com.google.mlkit	object-detection-custom	17.0.0	01/25/2022
com.google.mlkit	playstore-dynamic-feature-support	16.0.0-beta2	01/25/2022
com.google.mlkit	pose-detection	18.0.0-beta2	02/22/2022
com.google.mlkit	pose-detection-accurate	18.0.0-beta2	02/22/2022
com.google.mlkit	segmentation-selfie	16.0.0-beta4	01/25/2022
com.google.mlkit	smart-reply	17.0.1	04/26/2022
com.google.mlkit	text-recognition	16.0.0-beta4	04/26/2022
com.google.mlkit	text-recognition-chinese	16.0.0-beta4	04/26/2022
com.google.mlkit	text-recognition-devanagari	16.0.0-beta4	04/26/2022
com.google.mlkit	text-recognition-japanese	16.0.0-beta4	04/26/2022
com.google.mlkit	text-recognition-korean	16.0.0-beta4	04/26/2022
com.google.mlkit	translate	17.0.0	01/25/2022

Release History

May 10, 2022

This release includes new features.

New features

• Added new Google Code Scanner API on Android. The public beta of the Google code scanner provides a complete solution for scanning codes without requiring your app to request camera permission, while preserving user privacy. This is accomplished by delegating the task of scanning the code to Google Play services and returning only the scan results to your app.

April 26, 2022

This release includes improvements.

Improvements

• Improved Smart Reply model to better detect sensitive topics.
• On Android, upgraded all language models in the bundled version of the Text Recognition v2 API. Improved performance and added all-orientation support for Latin.
• On iOS, raised the minimum supported version of Xcode to 13.2.1.

February 22, 2022

This release includes new features and improvements.

New features

• On Android, added GPU acceleration support to Pose Detection.

Improvements

• On Android, upgraded models in the bundled version of the Face Detection API. Improved recall, latency for accurate mode and reduced the apk size impact from 11.6 MB to 6.9 MB.
• On iOS, support ECI code for CP437 text encoding in Barcode Scanning API.

January 25, 2022

This release includes internal improvements and bug fixes on Android.

Improvements

• Strict nullability annotations (@NonNull and @Nullable) are being used where previously @RecentlyNonNull and @RecentlyNullable were used. With this change, what once caused a nullness warning will now cause an error when building Kotlin code or using Java null checking frameworks.
• On Android, support ECI code for CP437 text encoding in Barcode Scanning API. .

Bug fixes

• Fixed a Barcode Scanning issue that code 128B and code 128C raw value has an additional character.

December 14, 2021

This release includes internal improvements and bug fixes.

Bug fixes

• Fixed an issue in Text Recognition v2 preventing text recognition results from returning recognized languages correctly.
• On Android, fixed an issue in Language Identification where the app may crash when upgrading to AGP 7.0.0 or later.
• On Android, fixed an issue in Pose Detection where the local reference table could overflow.
• On Android, fixed an issue in Digital Ink Recognition and Entity Extraction where a runtime exception could be thrown when the app targets API level 31 and runs on Android S.
• On iOS, return nil for rawValue when the barcode is not encoded in UTF-8 format to match API reference.

September 28, 2021

This release includes new features, improvements and bug fixes.

New features

• On Android, added unbundled version of Custom Image Labeling API, backed by Google Play Services, to reduce the impact on apk size from 5.5 MB to 600 KB.
• On iOS, added recognizedLanguages property to TextElement for Text Recognition v2.

Improvements

• On Android, enabled NNAPI(TPU) acceleration for Object Detection & Tracking with Base models and Pose Detection on Pixel 2021 devices.
• On iOS, raised the minimum supported version of Xcode to 12.5.1.
• On iOS, upgraded model for Language Identification with improved quality.

Bug fixes

• On Android, fixed an issue in Language Identification where the SDK could return an unexpected unknown tag.
• On iOS, fixed a bitcode recompilation issue when generating the .ipa file.

August 11, 2021

This release includes new features and improvements.

New features

• Added Text Recognition v2, which adds support for Chinese, Devanagari, Japanese and Korean scripts in addition to Latin and greatly increases range of supported languages. This API is offered as a bundled library and includes improved ML-based block/paragraph detection and updated models to boost recognition accuracy.
• On Android, added unbundled version of Language Identification, backed by Google Play Services, to reduce the impact on apk size by about 500KB.

Improvements

• Upgraded the bundled version of the Barcode Scanning API. Improved recall, long tail latency, low-quality image tolerance and bounding box stability.
• Upgraded the Pose Detection API for better accuracy. Also, added the experimental Z coordinates for facial landmarks.
• On Android, updated models for Language Identification, to improve accuracy.

Breaking changes

• On Android, there are two breaking changes for Text Recognition:
  • The deprecated TextRecognition.getClient() method is removed. Please use TextRecognition.getClient(TextRecognizerOptions.DEFAULT_OPTIONS) instead.
  • TextRecognizerOptions is moved from com.google.mlkit.vision.text to com.google.mlkit.vision.text.latin.
• On Android, for Barcode Scanning and Language Identification, the major version bumps are only due to breaking changes on non-public APIs. No code change is required when upgrading their dependency versions in your app.
• On Android, ML Kit no longer supports API levels 16, 17 and 18, due to extremely low use. API level 19 (KitKat) is now the minimum level supported.

Note

• On iOS, to keep your project on the v1 version of Text Recognition make sure to pin to version 2.2.0 or lower in your podfile. For example, adding pod 'GoogleMLKit/TextRecognition','2.2.0' to the pod file will pin your app to the latest v1 version of this feature.

June 29, 2021

This release includes new features and improvements.

New features

• Added support for com.google.android.odml.image.MlImage on Android and MLImage on iOS, a common image container (Beta) for on-device machine learning frameworks.

Improvements

• Upgraded the entity extraction models with improved quality for both Android and iOS.
• On Android, improved Pose Detection performance for certain CPU specs.
• On Android, fixed okhttp3 incompatible issue for Entity Extraction and Digital Ink Recognition APIs.
• On iOS, raised the minimum supported version of Xcode to 12.4.
• On iOS, upgraded the version range of GoogleDataTransport dependency in MLKitCommon from 8.x to 9.x and the version range of FirebaseCore dependency in MLKitLinkFirebase from 7.x to 8.x, to make ML Kit compatible with Firebase 8.x. Migrated MLKitLinkFirebase from FirebaseInstanceID to FirebaseInstallations.
• Breaking change: On iOS, removed label support for the deprecated remote AutoML-trained custom models without labels embedded in the model metadata. Please update such remote models hosted on Firebase Console with ones containing labels in its metadata.

June 02, 2021

This patch release fixed a bug in the com.google.mlkit:face-detection library.

May 25, 2021

This release includes new feature and improvements.

Improvements

• On Android, introduced a playstore dynamic feature support library for features depending on ML Kit and the ML Kit dependencies to be put in a dynamic feature module when app is built with play store app bundle.

• On Android, deprecated TextRecognition#getClient() API in Text Recognition SDK.

May 04, 2021

This patch release fixes a crash regression in the Custom Image Labeling API.

Bug fixes

• On Android, fixed a crash issue in Custom Image Labeling API when the model is downloaded from Firebase.

April 27, 2021

This release includes new feature and improvements.

Improvements

• On Android, introduced a camera helper library for MLKit vision features to be used with Camera output directly in streaming mode. It uses CameraX internally.

• On Android, enabled NNAPI acceleration for Object Detection & Tracking with Base models on more devices which provides faster model inference and saves energy.

Bug fixes

• On Android, fixed an issue in FaceDetection API when the InputImage is constructed by bitmap with a non-zero rotationDegrees.

March 30, 2021

This release includes improvements and bug fixes.

Improvements

• Added new models for Devanagari script languages for Digital Ink Recognition APIs.
• On Android, updated version of external dependencies for Face Detection APIs and Image Labeling APIs.

Bug fixes

• On Android, added Network permission to Translate API so app doesn't need to declare it.

March 02, 2021

This release includes new features and improvements.

New features

• Added Selfie Segmentation API, a CPU-based solution for developers to easily separate the background from users within a scene and focus on what matters.

Improvements

• Raised the minimum supported version of Xcode to 12.2 in iOS.

Feburary 5, 2021

This patch release fixes a regression bug in the Face Detection APIs.

Improvements and bug fixes

• There was a regression bug introduced in Face Detection versions 16.1.3 (thin) and 16.0.4 (bundled) which resulted in poor face detection. This patch release fixes the bug.

January 26, 2021

This release includes new features, improvements and bug fixes.

New features

• New Z Coordinate support in the Pose Detection API.

Improvements and bug fixes

• The Pose Detection API recognizes more poses, such as when a user is directly facing the camera while doing squats and pushups.
• The Pose Detection API base and accurate models are significantly smaller. This change does not impact the quality of the models.
• The Android Vision APIs have been updated. The size of the unbundled variants in the pull-in SDKs backed by Google Play Services is reduced by approximately 16% (~160KB).

December 8, 2020

This release includes new features, improvements and bug fixes.

New features

• ML Kit is no longer in Beta. All of ML Kit's APIs (excluding Pose Detection and Entity Extraction) have graduated to General Availability status.

• Added the Entity Extraction API (Beta). This API allows you to detect and locate entities (e.g. addresses, date/time etc.) from raw text. It supports 11 entity types and 15 languages, using the same technology that powers smart text selection in Android Q+.

  Entity Extraction works fully offline (aside from a one-time download of selected language model, which is ~700KB), and supports both Android and IOS.

Improvements and bug fixes

• Breaking change: The AutoML image labeling API is removed. Instead, custom image classification models trained using AutoML Vision Edge are now supported by the Custom Image Labeling and Custom Object Detection & Tracking APIs directly.
• The Custom Image Labeling and Custom Object Detection & Tracking now support remote models that are hosted in Firebase.
  • Added RemoteModelSource, FirebaseModelSource, and CustomRemoteModel classes to support custom models hosted remotely on Firebase server. Also added the support for creating a LocalModel instance from a manifest path of a local custom model trained by AutoML Vision Edge.
  • For the custom image labeling API, added the remote custom model support to the CustomImageLabelerOptions class.
  • For the custom object detection API, added the remote custom model support to the CustomObjectDetectorOptions class.
• On Android, significantly reduced the SDK size for Vision APIs. For APIs with unbundled variants (backed by Google Play Services) the pull-in SDK's size reduces by ~900KB (~65% of original size). For all other Vision APIs, the pull-in SDK's size reduces by ~600KB. The exception is Digital Ink Recognition, for which the SDK size remains the same.
• On Android, enabled NNAPI acceleration for Object Detection & Tracking with Base models in STREAM_MODE on selected devices and OSes (Pixel 3, 3XL, 4, 4XL with Android Q or R at the moment), which provides faster model inference and saves energy.
• On Android, fixed an issue in Barcode Scanning API where it did not work on some devices running with camera-camera2:1.0.0-beta08 version or later.

November 2, 2020

This release includes improvements and bug fixes.

iOS: GoogleMLKit/LinkFirebase 0.64.1

Improvements and bug fixes

• Upgraded the version range of FirebaseCore dependency in MLKitLinkFirebase from 6.x to 7.x, FirebaseInstanceID dependency from 4.x to 7.x.

October 29, 2020

This release includes improvements and bug fixes.

iOS: GoogleMLKit/Common 0.64.2

Improvements and bug fixes

• Upgraded the version range of GoogleUtilities/UserDefaults dependency in MLKitCommon from 6.x to 7.x.

October 26, 2020

This release includes improvements and bug fixes.

iOS: GoogleMLKit/Common 0.64.1

Improvements and bug fixes

• Upgraded the version range of GoogleDataTransport dependency in MLKitCommon from 7.x to 8.x.

September 29, 2020

This release includes new features, improvements and bug fixes.

iOS API Version: 0.64.0

New features

• On Android, added a “thin” variant of Image Labeling API, backed by Google Play Services. Reduces size impact to your app by ~3 MB compared to the bundled variant.

Improvements and bug fixes

• Breaking change: Split Pose Detection into two libraries for size optimization. One library only bundles the fast models, while the other only bundles the accurate models. This helps save the application size, if you only use one set of the models. Same change happened to iOS: The PoseDetection pod now contains only the fast models. To use the accurate models, use the PoseDetectionAccurate pod instead.
• On iOS, fixed a memory leak which occurred during pose detector deallocation.

August 25, 2020

This release includes new features, improvements and bug fixes.

iOS API Version: 0.63.0

New features

• Added the Pose Detection API, a CPU-based, lightweight and versatile solution for developers to track a user’s physical actions in real time within their apps. Pose Detection offers a 33 point skeletal match (with hands and feet tracking) of a users body in real time and has been tuned to work well for complex athletic use cases.

  Pose detection also works fully offline, comes with two operating modes (fast and accurate), and supports both Android and iOS.

Improvements and bug fixes

• Added support to create a LocalModel with a URI on Android.
• New Latin script recognizers for Filipino (fil/tl), Uzbek (uz-Latn), and Berber (ber-Latn) in Digital Ink Recognition.
• Improved models for Cyrillic script languages in Digital Ink Recognition.
• Changed the Hebrew language code from iw to he in Language Identification on Android since iw is deprecated in BCP-47.

July 28, 2020

This release includes new features, improvements, and bug fixes.

iOS API Version: 0.62.0

New features

• Added a new Digital Ink Recognition API that recognizes text and shapes handwritten on a digital surface (e.g. touch screens). Supports 300+ languages, as well as emoji and autodraw. It is the same technology that powers handwriting recognition layouts in Gboard, the Google Translate apps and the Quick, Draw! game.

  Digital Ink Recognition works fully offline (aside from a one-time download of language packs), and supports both Android and iOS.

• On Android, added support for specifying your own custom Executor for running expensive tasks like model loading and inference to Image Labeling and Object Detection and Tracking. With this release, all ML Kit APIs now support custom executors.

  By default, ML Kit uses a highly optimized, internally managed thread pool to run background tasks. This API can help with specialized use cases where developers want to keep full control over the threads in their app instead.

Improvements and bug fixes

• On Android, for the Barcode Scanning API, fixed rotation handling of the input image.

July 15, 2020

This release includes improvements and bug fixes.

iOS: GoogleMLKit/Common 0.61.1

Improvements and bug fixes

• Upgraded the version range of GoogleDataTransport dependency in MLKitCommon from 3.2 or higher to 7.0 or higher.
• Removed the GoogleDataTransportCCTSupport dependency from MLKitCommon.

July 1, 2020

This release includes new features, improvements and bug fixes.

iOS API Version: 0.61.0

New features

• The Image Labeling and Object Detection and Tracking APIs now support float-based custom models.

• On Android, added support for specifying your own custom Executor for running expensive tasks like model loading and inference to all APIs except Image Labeling and Object Detection and Tracking.

  By default, ML Kit uses a highly optimized, internally managed thread pool to run background tasks. This API can help with specialized use cases where developers want to keep full control over the threads in their app instead.

Improvements and bug fixes

• Breaking change: On iOS, for the Barcode Scanning API, fixed a typo in BarcodePersonName and renamed its pronounciation property to pronunciation.
• Fixed an issue on iOS in the Translate and Smart Reply APIs where the app would crash if the device locale does not specify a region (e.g. "en" instead of "en-US").

June 3, 2020

Android: 16.0.0 / iOS: 0.60.0

This is the first release of ML Kit as a standalone SDK, independent from Firebase. This SDK offers all the on-device APIs that were previously offered through the ML Kit for Firebase SDK. For more information on this change and instructions on migrating your existing apps, please follow our migration guide.

This release includes new features, improvements and bug fixes.

New features

• New Custom Image Labeling and Custom Object Detection and Tracking APIs, making it easy to use custom image classification models (pre-trained, or newly trained from a range of sources) in your apps and build real-time interactive experiences for users.

Improvements and bug fixes

• On Android, included support for Face Contours in “thin” variant of Face Detection API, backed by Google Play Services. Reduces size impact to your app by ~20.3 MB compared to bundled variant.
• On Android, improved CameraX and Camera2 performance for Barcode Scanning and Face Detection APIs by moving image processing code from Java to native code.
• Android Jetpack Lifecycle support has been added to all APIs. Developers can use addObserver to automatically manage the initiation and teardown of ML Kit APIs as the app goes through screen rotation or closure by the user / system. This makes CameraX integration easier.
• Text Recognition API: the most prevalent recognized language is now provided
• Face Detection API: the Euler X angle of a face is now provided.
• Barcode Scanning API [bundled]: added support for broken PDF417 start/stop pattern detection, improving recall by 10%.
• Object Detection and Tracking API: updated localizer model, improving average precision and reducing footprint by ~700KB.

Legacy releases

Changes prior to June 3, 2020 can be found in the Firebase release notes.