Releases · infinitered/react-native-mlkit

07 Feb 23:01

github-actions

@infinitered/react-native-mlkit-object-detection@3.0.0

3b07ad6

@infinitered/react-native-mlkit-object-detection@3.0.0 Latest

Latest

Major Changes

83b7e6e: Standardize API patterns and coordinate structures across ML Kit modules

Separates model operations into three hooks with simpler APIs
1. loading the models, (useObjectDetectionModels, useImageLabelingModels)
2. initializing the provider (useObjectDetectionProvider, useImageLabelingProvider)
3. accessing models for inference, (useObjectDetector, useImageLabeling)
Implements consistent naming patterns to make the APIs more legible
- Removes "RNMLKit" prefix from non-native types
- Use specific names for hooks (useImageLabelingModels instead of useModels)
- Model configs are now Configs, instead of AssetRecords
Moves base types into the core package to ensure consistency
Fixes an issue with bounding box placement on portrait / rotated images on iOS
Improves error handling and state management
Updates documentation to match the new API

Breaking Changes

Image Labeling

Renamed useModels to useImageLabelingModels for clarity
Renamed useImageLabeler to useImageLabeling
Introduced new useImageLabelingProvider hook for cleaner context management
Added type-safe configurations with ImageLabelingConfig
Renamed model context provider from ObjectDetectionModelContextProvider to ImageLabelingModelProvider

Here's how to update your app:

Fetching the provider

- const MODELS: AssetRecord = {
+ const MODELS: ImageLabelingConfig = {
 nsfwDetector: {
   model: require("./assets/models/nsfw-detector.tflite"),
   options: {
     maxResultCount: 5,
     confidenceThreshold: 0.5,
   }
 },
};

function App() {
- const { ObjectDetectionModelContextProvider } = useModels(MODELS)
+ const models = useImageLabelingModels(MODELS)
+ const { ImageLabelingModelProvider } = useImageLabelingProvider(models)

 return (
-   <ObjectDetectionModelContextProvider>
+   <ImageLabelingModelProvider>
     {/* Rest of your app */}
-   </ObjectDetectionModelContextProvider>
+   </ImageLabelingModelProvider>
 )
}

Using the model

- const model = useImageLabeler("nsfwDetector")
+ const detector = useImageLabeling("nsfwDetector")

const labels = await detector.classifyImage(imagePath)

Object Detection

useObjectDetectionModels now requires an assets parameter
useObjectDetector is now useObjectDetection
Introduced new useObjectDetectionProvider hook for context management
Renamed and standardized type definitions:
- RNMLKitObjectDetectionObject → ObjectDetectionObject
- RNMLKitObjectDetectorOptions → ObjectDetectorOptions
- RNMLKitCustomObjectDetectorOptions → CustomObjectDetectorOptions
Added new types: ObjectDetectionModelInfo, ObjectDetectionConfig, ObjectDetectionModels
Moved model configuration to typed asset records
Default model now included in models type union

Here's how to update your app:

Fetching the provider

- const MODELS: AssetRecord = {
+ const MODELS: ObjectDetectionConfig = {
  birdDetector: {
    model: require("./assets/models/bird-detector.tflite"),
    options: {
      shouldEnableClassification: false,
      shouldEnableMultipleObjects: false,
    }
  },
};

function App() {

- const { ObjectDetectionModelContextProvider } = useObjectDetectionModels({
-    assets: MODELS,
-    loadDefaultModel: true,
-    defaultModelOptions: DEFAULT_MODEL_OPTIONS,
-   })

+ const models = useObjectDetectionModels({
+   assets: MODELS,
+   loadDefaultModel: true,
+   defaultModelOptions: DEFAULT_MODEL_OPTIONS,
+ })
+
+ const { ObjectDetectionProvider } = useObjectDetectionProvider(models)

  return (
-    <ObjectDetectionModelContextProvider>
+    <ObjectDetectionProvider>
       {/* Rest of your app */}
-   </ObjectDetectionModelContextProvider>
+   </ObjectDetectionProvider>
)
}

Using the model

- const {models: {birdDetector} = useObjectDetectionModels({
-  assets: MODELS,
-  loadDefaultModel: true,
-   defaultModelOptions: DEFAULT_MODEL_OPTIONS,
- })
-
+ const birdDetector = useObjectDetection("birdDetector")

const objects = birdDetector.detectObjects(imagePath)

Face Detection

Changed option naming conventions to match ML Kit SDK patterns:
detectLandmarks → landmarkMode
runClassifications → classificationMode
Changed default performanceMode from accurate to fast
Renamed hook from useFaceDetector to useFaceDetection
Renamed context provider from RNMLKitFaceDetectionContextProvider to FaceDetectionProvider
Added comprehensive error handling
Added new state management with FaceDetectionState type

Here's how to update your app:

Using the detector

const options = {
-  detectLandmarks: true,
+  landmarkMode: true,
-  runClassifications: true,
+  classificationMode: true,
}

Using the provider

- import { RNMLKitFaceDetectionContextProvider } from "@infinitered/react-native-mlkit-face-detection"
+ import { FaceDetectionProvider } from "@infinitered/react-native-mlkit-face-detection"

function App() {
 return (
-   <RNMLKitFaceDetectionContextProvider>
+   <FaceDetectionProvider>
     {/* Rest of your app */}
-   </RNMLKitFaceDetectionContextProvider>
+   </FaceDetectionProvider>
 )
}

Using the hooks

- const detector = useFaceDetector()
+ const detector = useFaceDetection()

// useFacesInPhoto remains unchanged
const { faces, status, error } = useFacesInPhoto(imageUri)

Core Module

Introduced shared TypeScript interfaces:
- ModelInfo<T>
- AssetRecord<T>
Standardized frame coordinate structure
Implemented consistent type patterns

b668ab0: Upgrade to expo 52

Minor Changes

b668ab0: align podspec platform requirements with expo version

Patch Changes

Updated dependencies [b668ab0]
Updated dependencies [83b7e6e]
Updated dependencies [b668ab0]
Updated dependencies [213f085]
- @infinitered/react-native-mlkit-core@3.0.0

Assets 2

07 Feb 23:01

github-actions

@infinitered/react-native-mlkit-image-labeling@3.0.0

3b07ad6

@infinitered/react-native-mlkit-image-labeling@3.0.0

Major Changes

83b7e6e: Standardize API patterns and coordinate structures across ML Kit modules

Separates model operations into three hooks with simpler APIs
1. loading the models, (useObjectDetectionModels, useImageLabelingModels)
2. initializing the provider (useObjectDetectionProvider, useImageLabelingProvider)
3. accessing models for inference, (useObjectDetector, useImageLabeling)
Implements consistent naming patterns to make the APIs more legible
- Removes "RNMLKit" prefix from non-native types
- Use specific names for hooks (useImageLabelingModels instead of useModels)
- Model configs are now Configs, instead of AssetRecords
Moves base types into the core package to ensure consistency
Fixes an issue with bounding box placement on portrait / rotated images on iOS
Improves error handling and state management
Updates documentation to match the new API

Breaking Changes

Image Labeling

Renamed useModels to useImageLabelingModels for clarity
Renamed useImageLabeler to useImageLabeling
Introduced new useImageLabelingProvider hook for cleaner context management
Added type-safe configurations with ImageLabelingConfig
Renamed model context provider from ObjectDetectionModelContextProvider to ImageLabelingModelProvider

Here's how to update your app:

Fetching the provider

- const MODELS: AssetRecord = {
+ const MODELS: ImageLabelingConfig = {
 nsfwDetector: {
   model: require("./assets/models/nsfw-detector.tflite"),
   options: {
     maxResultCount: 5,
     confidenceThreshold: 0.5,
   }
 },
};

function App() {
- const { ObjectDetectionModelContextProvider } = useModels(MODELS)
+ const models = useImageLabelingModels(MODELS)
+ const { ImageLabelingModelProvider } = useImageLabelingProvider(models)

 return (
-   <ObjectDetectionModelContextProvider>
+   <ImageLabelingModelProvider>
     {/* Rest of your app */}
-   </ObjectDetectionModelContextProvider>
+   </ImageLabelingModelProvider>
 )
}

Using the model

- const model = useImageLabeler("nsfwDetector")
+ const detector = useImageLabeling("nsfwDetector")

const labels = await detector.classifyImage(imagePath)

Object Detection

useObjectDetectionModels now requires an assets parameter
useObjectDetector is now useObjectDetection
Introduced new useObjectDetectionProvider hook for context management
Renamed and standardized type definitions:
- RNMLKitObjectDetectionObject → ObjectDetectionObject
- RNMLKitObjectDetectorOptions → ObjectDetectorOptions
- RNMLKitCustomObjectDetectorOptions → CustomObjectDetectorOptions
Added new types: ObjectDetectionModelInfo, ObjectDetectionConfig, ObjectDetectionModels
Moved model configuration to typed asset records
Default model now included in models type union

Here's how to update your app:

Fetching the provider

- const MODELS: AssetRecord = {
+ const MODELS: ObjectDetectionConfig = {
  birdDetector: {
    model: require("./assets/models/bird-detector.tflite"),
    options: {
      shouldEnableClassification: false,
      shouldEnableMultipleObjects: false,
    }
  },
};

function App() {

- const { ObjectDetectionModelContextProvider } = useObjectDetectionModels({
-    assets: MODELS,
-    loadDefaultModel: true,
-    defaultModelOptions: DEFAULT_MODEL_OPTIONS,
-   })

+ const models = useObjectDetectionModels({
+   assets: MODELS,
+   loadDefaultModel: true,
+   defaultModelOptions: DEFAULT_MODEL_OPTIONS,
+ })
+
+ const { ObjectDetectionProvider } = useObjectDetectionProvider(models)

  return (
-    <ObjectDetectionModelContextProvider>
+    <ObjectDetectionProvider>
       {/* Rest of your app */}
-   </ObjectDetectionModelContextProvider>
+   </ObjectDetectionProvider>
)
}

Using the model

- const {models: {birdDetector} = useObjectDetectionModels({
-  assets: MODELS,
-  loadDefaultModel: true,
-   defaultModelOptions: DEFAULT_MODEL_OPTIONS,
- })
-
+ const birdDetector = useObjectDetection("birdDetector")

const objects = birdDetector.detectObjects(imagePath)

Face Detection

Changed option naming conventions to match ML Kit SDK patterns:
detectLandmarks → landmarkMode
runClassifications → classificationMode
Changed default performanceMode from accurate to fast
Renamed hook from useFaceDetector to useFaceDetection
Renamed context provider from RNMLKitFaceDetectionContextProvider to FaceDetectionProvider
Added comprehensive error handling
Added new state management with FaceDetectionState type

Here's how to update your app:

Using the detector

const options = {
-  detectLandmarks: true,
+  landmarkMode: true,
-  runClassifications: true,
+  classificationMode: true,
}

Using the provider

- import { RNMLKitFaceDetectionContextProvider } from "@infinitered/react-native-mlkit-face-detection"
+ import { FaceDetectionProvider } from "@infinitered/react-native-mlkit-face-detection"

function App() {
 return (
-   <RNMLKitFaceDetectionContextProvider>
+   <FaceDetectionProvider>
     {/* Rest of your app */}
-   </RNMLKitFaceDetectionContextProvider>
+   </FaceDetectionProvider>
 )
}

Using the hooks

- const detector = useFaceDetector()
+ const detector = useFaceDetection()

// useFacesInPhoto remains unchanged
const { faces, status, error } = useFacesInPhoto(imageUri)

Core Module

Introduced shared TypeScript interfaces:
- ModelInfo<T>
- AssetRecord<T>
Standardized frame coordinate structure
Implemented consistent type patterns

b668ab0: Upgrade to expo 52

Minor Changes

b668ab0: align podspec platform requirements with expo version

Patch Changes

Updated dependencies [b668ab0]
Updated dependencies [83b7e6e]
Updated dependencies [b668ab0]
Updated dependencies [213f085]
- @infinitered/react-native-mlkit-core@3.0.0

Assets 2

07 Feb 23:01

github-actions

@infinitered/react-native-mlkit-face-detection@3.0.0

3b07ad6

@infinitered/react-native-mlkit-face-detection@3.0.0

Major Changes

b668ab0: Upgrade to expo 52

Face Detection

Changed option naming conventions to match ML Kit SDK patterns:
detectLandmarks → landmarkMode
runClassifications → classificationMode
Changed default performanceMode from accurate to fast
Renamed hook from useFaceDetector to useFaceDetection
Renamed context provider from RNMLKitFaceDetectionContextProvider to FaceDetectionProvider
Added comprehensive error handling
Added new state management with FaceDetectionState type

Here's how to update your app:

Using the detector

const options = {
-  detectLandmarks: true,
+  landmarkMode: true,
-  runClassifications: true,
+  classificationMode: true,
}

Using the provider

- import { RNMLKitFaceDetectionContextProvider } from "@infinitered/react-native-mlkit-face-detection"
+ import { FaceDetectionProvider } from "@infinitered/react-native-mlkit-face-detection"

function App() {
 return (
-   <RNMLKitFaceDetectionContextProvider>
+   <FaceDetectionProvider>
     {/* Rest of your app */}
-   </RNMLKitFaceDetectionContextProvider>
+   </FaceDetectionProvider>
 )
}

Using the hooks

- const detector = useFaceDetector()
+ const detector = useFaceDetection()

// useFacesInPhoto remains unchanged
const { faces, status, error } = useFacesInPhoto(imageUri)

Minor Changes

b668ab0: align podspec platform requirements with expo version

Patch Changes

Updated dependencies [b668ab0]
Updated dependencies [83b7e6e]
Updated dependencies [b668ab0]
Updated dependencies [213f085]
@infinitered/react-native-mlkit-core@3.0.0

Assets 2

07 Feb 23:01

github-actions

@infinitered/react-native-mlkit-document-scanner@3.0.0

3b07ad6

@infinitered/react-native-mlkit-document-scanner@3.0.0

Major Changes

b668ab0: Upgrade to expo 52

Patch Changes

Updated dependencies [b668ab0]
Updated dependencies [83b7e6e]
Updated dependencies [b668ab0]
Updated dependencies [213f085]
- @infinitered/react-native-mlkit-core@3.0.0

Assets 2

07 Feb 23:01

github-actions

@infinitered/react-native-mlkit-core@3.0.0

3b07ad6

@infinitered/react-native-mlkit-core@3.0.0

Major Changes

83b7e6e: Standardize API patterns and coordinate structures across ML Kit modules

Separates model operations into three hooks with simpler APIs
1. loading the models, (useObjectDetectionModels, useImageLabelingModels)
2. initializing the provider (useObjectDetectionProvider, useImageLabelingProvider)
3. accessing models for inference, (useObjectDetector, useImageLabeling)
Implements consistent naming patterns to make the APIs more legible
- Removes "RNMLKit" prefix from non-native types
- Use specific names for hooks (useImageLabelingModels instead of useModels)
- Model configs are now Configs, instead of AssetRecords
Moves base types into the core package to ensure consistency
Fixes an issue with bounding box placement on portrait / rotated images on iOS
Improves error handling and state management
Updates documentation to match the new API

Breaking Changes

Image Labeling

Renamed useModels to useImageLabelingModels for clarity
Renamed useImageLabeler to useImageLabeling
Introduced new useImageLabelingProvider hook for cleaner context management
Added type-safe configurations with ImageLabelingConfig
Renamed model context provider from ObjectDetectionModelContextProvider to ImageLabelingModelProvider

Here's how to update your app:

Fetching the provider

- const MODELS: AssetRecord = {
+ const MODELS: ImageLabelingConfig = {
 nsfwDetector: {
   model: require("./assets/models/nsfw-detector.tflite"),
   options: {
     maxResultCount: 5,
     confidenceThreshold: 0.5,
   }
 },
};

function App() {
- const { ObjectDetectionModelContextProvider } = useModels(MODELS)
+ const models = useImageLabelingModels(MODELS)
+ const { ImageLabelingModelProvider } = useImageLabelingProvider(models)

 return (
-   <ObjectDetectionModelContextProvider>
+   <ImageLabelingModelProvider>
     {/* Rest of your app */}
-   </ObjectDetectionModelContextProvider>
+   </ImageLabelingModelProvider>
 )
}

Using the model

- const model = useImageLabeler("nsfwDetector")
+ const detector = useImageLabeling("nsfwDetector")

const labels = await detector.classifyImage(imagePath)

Object Detection

useObjectDetectionModels now requires an assets parameter
useObjectDetector is now useObjectDetection
Introduced new useObjectDetectionProvider hook for context management
Renamed and standardized type definitions:
- RNMLKitObjectDetectionObject → ObjectDetectionObject
- RNMLKitObjectDetectorOptions → ObjectDetectorOptions
- RNMLKitCustomObjectDetectorOptions → CustomObjectDetectorOptions
Added new types: ObjectDetectionModelInfo, ObjectDetectionConfig, ObjectDetectionModels
Moved model configuration to typed asset records
Default model now included in models type union

Here's how to update your app:

Fetching the provider

- const MODELS: AssetRecord = {
+ const MODELS: ObjectDetectionConfig = {
  birdDetector: {
    model: require("./assets/models/bird-detector.tflite"),
    options: {
      shouldEnableClassification: false,
      shouldEnableMultipleObjects: false,
    }
  },
};

function App() {

- const { ObjectDetectionModelContextProvider } = useObjectDetectionModels({
-    assets: MODELS,
-    loadDefaultModel: true,
-    defaultModelOptions: DEFAULT_MODEL_OPTIONS,
-   })

+ const models = useObjectDetectionModels({
+   assets: MODELS,
+   loadDefaultModel: true,
+   defaultModelOptions: DEFAULT_MODEL_OPTIONS,
+ })
+
+ const { ObjectDetectionProvider } = useObjectDetectionProvider(models)

  return (
-    <ObjectDetectionModelContextProvider>
+    <ObjectDetectionProvider>
       {/* Rest of your app */}
-   </ObjectDetectionModelContextProvider>
+   </ObjectDetectionProvider>
)
}

Using the model

- const {models: {birdDetector} = useObjectDetectionModels({
-  assets: MODELS,
-  loadDefaultModel: true,
-   defaultModelOptions: DEFAULT_MODEL_OPTIONS,
- })
-
+ const birdDetector = useObjectDetection("birdDetector")

const objects = birdDetector.detectObjects(imagePath)

Face Detection

Changed option naming conventions to match ML Kit SDK patterns:
detectLandmarks → landmarkMode
runClassifications → classificationMode
Changed default performanceMode from accurate to fast
Renamed hook from useFaceDetector to useFaceDetection
Renamed context provider from RNMLKitFaceDetectionContextProvider to FaceDetectionProvider
Added comprehensive error handling
Added new state management with FaceDetectionState type

Here's how to update your app:

Using the detector

const options = {
-  detectLandmarks: true,
+  landmarkMode: true,
-  runClassifications: true,
+  classificationMode: true,
}

Using the provider

- import { RNMLKitFaceDetectionContextProvider } from "@infinitered/react-native-mlkit-face-detection"
+ import { FaceDetectionProvider } from "@infinitered/react-native-mlkit-face-detection"

function App() {
 return (
-   <RNMLKitFaceDetectionContextProvider>
+   <FaceDetectionProvider>
     {/* Rest of your app */}
-   </RNMLKitFaceDetectionContextProvider>
+   </FaceDetectionProvider>
 )
}

Using the hooks

- const detector = useFaceDetector()
+ const detector = useFaceDetection()

// useFacesInPhoto remains unchanged
const { faces, status, error } = useFacesInPhoto(imageUri)