passive_liveness 0.0.1

Ultra-lightweight passive face liveness detection using LiteRT (TensorFlow Lite) for Flutter on iOS and Android.

passive_liveness

An ultra-lightweight, high-performance passive face anti-spoofing detection package for Flutter using LiteRT (TensorFlow Lite) edge inference on Android and iOS.

---

Features

• ⚡ High Performance Edge Inference: Powered by LiteRT (flutter_litert), achieving ~20–40ms inference time on mobile devices.
• 📷 Zero-Copy Camera Stream Processing: Preprocesses raw CameraImage byte buffers (NV21 / YUV420 on Android, BGRA8888 on iOS) directly to TFLite tensors without main-thread image decoding.
• 🖼️ Zero-Dependency Static Photo File Detection: Uses Flutter's built-in C++ Skia engine codecs (dart:ui) to decode static images (File or Uint8List) with 0 external image package dependencies.
• 📱 Rotation & Sensor Alignment: Closed-form pixel sampling matching OpenCV BORDER_REFLECT_101 supporting 0°, 90°, 180°, and 270° camera sensor rotation.
• 💡 Low-Light Shadow-Lift Compensation: Automatically un-clips shadow gradients in dark or underexposed room environments.

---

Installation

Add passive_liveness to your pubspec.yaml:

dependencies:
  passive_liveness: ^0.0.1

Or run:

flutter pub add passive_liveness

Ensure your pubspec.yaml bundles the bundled MiniFAS TFLite model asset:

flutter:
  assets:
    - packages/passive_liveness/assets/best_model.tflite

---

Usage

1. Initialize Detector Engine

import 'package:passive_liveness/passive_liveness.dart';

final detector = PassiveLivenessDetector();
await detector.initialize();

---

2. Detect Liveness from Camera Frame Stream (CameraImage)

import 'package:camera/camera.dart';
import 'package:passive_liveness/passive_liveness.dart';

void onFrameReceived(CameraImage cameraImage, FaceBoundingBox? boundingBox, int sensorRotation) async {
  // Convert CameraImage into LivenessImageBuffer
  final buffer = LivenessImageBuffer(
    width: cameraImage.width,
    height: cameraImage.height,
    format: cameraImage.format.group == ImageFormatGroup.bgra8888
        ? LivenessImageFormat.bgra8888
        : (cameraImage.planes.length == 1
            ? LivenessImageFormat.nv21
            : LivenessImageFormat.yuv420),
    planes: cameraImage.planes
        .map((p) => LivenessImagePlane(
              bytes: p.bytes,
              bytesPerRow: p.bytesPerRow,
              bytesPerPixel: p.bytesPerPixel,
            ))
        .toList(),
  );

  final LivenessResult result = await detector.detectLivenessFromBuffer(
    buffer,
    boundingBox: boundingBox, // From ML Kit face detection (in raw buffer coords)
    rotation: sensorRotation,  // e.g., 90 for iOS front camera
  );

  if (result.isReal) {
    print('Real human face detected! Score: ${result.realScore.toStringAsFixed(3)}');
  } else {
    print('Spoof face detected! Score: ${result.spoofScore.toStringAsFixed(3)}');
  }
}

---

3. Detect Liveness from Static Photo File (File / ImagePicker)

import 'dart:io';
import 'package:passive_liveness/passive_liveness.dart';

Future<void> checkPhotoLiveness(File imageFile, FaceBoundingBox? boundingBox) async {
  final LivenessResult result = await detector.detectLivenessFromImageFile(
    imageFile,
    boundingBox: boundingBox,
  );

  print('Is Real: ${result.isReal}');
  print('Real Score: ${result.realScore}');
  print('Inference Time: ${result.inferenceTime.inMilliseconds}ms');
}

---

4. Detect Liveness from Encoded Image Bytes (Uint8List)

import 'dart:typed_data';
import 'package:passive_liveness/passive_liveness.dart';

Future<void> checkBytesLiveness(Uint8List imageBytes) async {
  final LivenessResult result = await detector.detectLivenessFromImageBytes(
    imageBytes,
  );

  print('Is Real: ${result.isReal}');
}

---

Model Attribution & Credits

This package utilizes the MiniFASNet v2 SE passive face anti-spoofing model architecture, derived and converted from facenox/face-antispoof-onnx (MiniFASNetV2 SE trained with Fourier Transform frequency spectrum loss).

---

Author

Andika Tri Prasetya

• GitHub: github.com/andikatp

---

License

This project is licensed under the OSI-approved MIT License - see the LICENSE file for details.