camera_process #

On-device face detection and text recognition for Flutter, powered by Google ML Kit. Works on still images and on a live camera stream, fully offline, on both Android and iOS.

Donation #

If this package helps you, consider supporting its development:

Features #

• Runs entirely on-device — no network calls, no API keys.
• Detect faces with landmarks, contours, head angles, and smile / eyes-open probabilities.
• Recognize text as a hierarchy of blocks → lines → elements, each with bounding boxes.
• Feed it a file, a File, or raw camera bytes.

Requirements #

Android #

• minSdkVersion 21
• compileSdkVersion 34

iOS #

• Minimum deployment target: iOS 10.0
• Uses CocoaPods (see the Swift Package Manager note below)

Installation #

Add the dependency to your pubspec.yaml:

dependencies:
  camera_process: ^1.1.0

Then run:

flutter pub get

Permissions #

This plugin does not request permissions itself — declare the ones your app uses.

Android (android/app/src/main/AndroidManifest.xml):

<uses-permission android:name="android.permission.CAMERA" />

iOS (ios/Runner/Info.plist):

<key>NSCameraUsageDescription</key>
<string>This app needs camera access to detect faces and text.</string>
<key>NSPhotoLibraryUsageDescription</key>
<string>This app needs photo library access to process selected images.</string>

Usage #

Text recognition #

import 'package:camera_process/camera_process.dart';

final textDetector = CameraProcess.vision.textDetector();

final inputImage = InputImage.fromFilePath('/path/to/image.jpg');
final RecognisedText recognisedText = await textDetector.processImage(inputImage);

print(recognisedText.text);
for (final block in recognisedText.blocks) {
  for (final line in block.lines) {
    print('${line.text} @ ${line.rect}');
  }
}

// Release native resources when done.
await textDetector.close();

Face detection #

import 'package:camera_process/camera_process.dart';

final faceDetector = CameraProcess.vision.faceDetector(
  const FaceDetectorOptions(
    enableContours: true,
    enableClassification: true,
  ),
);

final inputImage = InputImage.fromFilePath('/path/to/image.jpg');
final List<Face> faces = await faceDetector.processImage(inputImage);

for (final face in faces) {
  print('Face at ${face.boundingBox}');
  print('Smiling: ${face.smilingProbability}');
}

await faceDetector.close();

Building an InputImage #

// From a file path
InputImage.fromFilePath('/path/to/image.jpg');

// From a dart:io File
InputImage.fromFile(file);

// From raw camera-stream bytes
InputImage.fromBytes(bytes: bytes, inputImageData: inputImageData);

For a complete real-time example (live camera stream + gallery), including drawing the results with a CustomPainter, see the example/ app.

Swift Package Manager #

Flutter's Swift Package Manager (SPM) support is still gated by this plugin's native dependency: Google ML Kit ships as CocoaPods only and has no official SPM distribution. Until Google publishes SPM artifacts, the iOS side of this plugin must be built with CocoaPods.

An SPM scaffold is included at ios/camera_process/Package.swift.template and the migration path is documented in ios/SPM.md so the package can move to SPM the moment the ML Kit dependency supports it.

Contributing #

Issues and pull requests are welcome on GitHub.

License #

Released under the MIT License.