Google Ml kit Plugin

Pub Version

Flutter plugin to use google's standalone ml kit for Android .

Currently supported api's

Support for other api's will be shortly added


Add this plugin as dependency in your pubspec.yaml.

  • In your project-level build.gradle file, make sure to include Google's Maven repository in both your buildscript and allprojects sections(for all api's).
  • The plugin has been written using bundled api models, this implies models will be bundled along with plugin and there is no need to implement any dependencies on your part and should work out of the box.
  • If you wish to reduce the apk size you may replace bundled model dependencies with model's provided within Google Play Service, to know more about this see the below links
    1. Image Labeling
    2. Barcode Scanning

First Create an InputImage

Prepare Input Image(image you want to process)

import 'package:google_ml_kit/google_ml_kit.dart';

final inputImage = InputImage.fromFilePath(filePath);

// Or you can prepare image form
//final inputImage = InputImage.fromFile(file);

// final inputImageData = InputImageData(
//     size: size of image,
//     rotation: roatation degree(0,90,180,270 supported),
//     inputImageFormat:InputImageFormat.NV21 (default format of image);
// var inputImage = InputImage.fromBytes(
//     bytes: await pickedFile.readAsBytes(), 
//     inputImageData: inputImageData);

To know more about formats of image.

Create an instance of detector

Call processImage() to obtain the result

Call close()

An example covering all the api's usage

Digital Ink reognition

Read to know how to imlpement Digital Ink Recognition

Pose Detection

-Googgle Play service model is not available for this api' so no extra implementation*

-Create PoseDetectorOptions

final options = PoseDetectorOptions(
        poseDetectionModel: PoseDetectionModel.BasePoseDetector,
        selectionType : LandmarkSelectionType.all,
        poseLandmarks:(list of poseaLndmarks you want); 
//or PoseDetectionModel.AccuratePoseDetector to use accurate pose detector

-Obtain `PoseDetector` instance.

Note: To obtain default poseDetector no options need to be specied. It gives all available landmarks using BasePoseDetector Model.

The same implies to other detectors as well

PoseDetector poseDetector = GoogleMlKit.instance
                               .poseDetector([PoseDetectorOptions options]);

-Call processImage(InputImage inputImage) to obtain the result. It returns


final landMarksMap = await poseDetector.processImage(inputImage);

Use the map to extract data. See this example to get better idea.

Image Labeling

In plugin's build.gradle. For latest version check Image Labeling

dependencies {
       // ...
// Use this dependency to use dynamically downloaded model in Google Play Service
      implementation ''

If you choose google service way.In app level buil.gradle add

<application ...>
          android:value="ica" />
      <!-- To use multiple models: android:value="ica,model2,model3" -->

The same implies for all other models as well

Create ImageLabelerOptions. This uses google's base model

final options =ImageLabelerOptions( confidenceThreshold = confidenceThreshold);
// Default =0.5
//lies between 0.0 to 1.0

To use custom tflite models

CustomImageLabelerOptions options = CustomImageLabelerOptions(
        customModel: CustomTrainedModel.asset 
       (or CustomTrainedModel.file),// To use files stored in device
        customModelPath: "file path");

To use autoMl vision models models

final options = AutoMlImageLabelerOptions(
      customTrainedModel: CustomTrainedModel.asset 
       (or CustomTrainedModel.file), 

Obtain ImageLabeler instance.

ImageLabeler imageLabeler = GoogleMlKit.instance.imageLabeler([options]);

call processImage() It returns List<ImageLabel>

final labels = await imageLabeler.processImage(inputImage);

To know more see this example

Barcode Scanner

In you app-level build.gradle. For latest version check Barcode Scanning

dependencies {
      // ...
// Use this dependency to use the dynamically downloaded model in Google Play Services
      implementation ''

Obtain BarcodeScanner instance.

BarcodeScanner barcodeScanner = GoogleMlKit.instance
                                           formats:(List of BarcodeFormats);

Supported BarcodeFormats .Access them using




call processImage() It returns List<Barcode>

final result = await barcodeScanner.processImage(inputImage);

To know more see this example


Incase of any errors open an issue.

Please make sure to update tests as appropriate.