capacitor-plugin-faceantispoofing

Capacitor plugin for Passive Liveness Detection and Face Anti-Spoofing using TensorFlow Lite models.

This plugin has been migrated from cordova-plugin-facespoofing.

Features

• Face Detection: Uses MTCNN (Multi-task Cascaded Convolutional Networks) for accurate face detection
• Anti-Spoofing Detection: Passive liveness detection using TensorFlow Lite models
• Blur Detection: Laplacian variance-based image quality assessment
• Cross-Platform: Support for both Android and iOS

Install

Using npm

npm install capacitor-plugin-faceantispoofing
npx cap sync

Using yarn

yarn add capacitor-plugin-faceantispoofing
npx cap sync

Usage

import { FaceAntiSpoofing } from 'capacitor-plugin-faceantispoofing';

// Detect face liveness from base64 image
const detectLiveness = async (base64Image: string) => {
  try {
    const result = await FaceAntiSpoofing.detect({
      image: base64Image  // Base64 encoded image (without data:image/...;base64, prefix)
    });

    console.log('Error:', result.error);
    console.log('Liveness:', result.liveness);
    console.log('Score:', result.score);
    console.log('Threshold:', result.threshold);
    console.log('Message:', result.message);

    if (result.liveness) {
      console.log('Real face detected!');
    } else {
      console.log('Spoof detected or no face found!');
    }
  } catch (error) {
    console.error('Detection failed:', error);
  }
};

API

detect(options)

Detects face liveness and performs anti-spoofing detection on the provided image.

| Param | Type | Description | | ------------ | ----------------------------- | -------------------------------------------------- | | options | <DetectOptions> | Options for the detection | | image | <string> | Base64 encoded image data or file URI |

Returns: Promise<DetectionResult>

DetectOptions

| Prop | Type | Description | | ------- | ------------------- | ---------------------------------------- | | image | string | Base64 encoded image (without data:image/...;base64, prefix) or file URI starting with file:// |

DetectionResult

| Prop | Type | Description | | --------- | ------------------- | ---------------------------------------- | | error | boolean | Whether an error occurred | | liveness | boolean | Whether the face is detected as real | | score | string | The liveness score | | threshold | string | The threshold used for classification | | message | string | Additional information message |

Detection Flow

1. Face Detection: Uses MTCNN (P-Net → R-Net → O-Net) to detect faces in the image
2. Face Alignment: Aligns the detected face based on facial landmarks
3. Blur Detection: Checks image clarity using Laplacian variance
4. Anti-Spoofing: If image is clear enough, runs the anti-spoofing model to classify as real or fake

Response Values

• No faces detected: error: true, message: "No faces detected"
• Image too blurry: error: false, liveness: false, score contains Laplacian value
• Real face detected: error: false, liveness: true, score < threshold
• Spoof detected: error: false, liveness: false, score >= threshold

Platform Support

| Platform | Supported | | -------- | --------- | | Android | ✅ Yes | | iOS | ✅ Yes | | Web | ❌ No* |

*Web is not supported as native TensorFlow Lite is required.

Dependencies

Android

• Google AI Edge LITErt (TensorFlow Lite) 1.4.0
• minSdkVersion: 24
• compileSdkVersion: 36

iOS

• TensorFlowLiteSwift ~ 2.17.0
• TensorFlowLiteC ~ 2.17.0
• iOS Deployment Target: 15.0

Model Files

This plugin includes the following TensorFlow Lite models:

• FaceAntiSpoofing.tflite - Anti-spoofing classification model
• pnet.tflite - MTCNN Proposal Network
• rnet.tflite - MTCNN Refine Network
• onet.tflite - MTCNN Output Network

Threshold Values

• Android: THRESHOLD = 0.5, LAPLACIAN_THRESHOLD = 0
• iOS: threshold = 0.5, laplacianThreshold = 0

Note: These values have been normalized from the original Cordova plugin.

Example Image Format

The plugin accepts images in the following formats:

1. Base64 without prefix: iVBORw0KGgoAAAANSUhEUgAA...
2. Base64 with data URI: data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAA...
3. File URI: file:///path/to/image.png

Error Handling

try {
  const result = await FaceAntiSpoofing.detect({ image: base64Data });

  if (result.error) {
    console.error('Detection error:', result.message);
    return;
  }

  if (result.liveness) {
    // Real face detected
  } else {
    // Spoof or no face detected
  }
} catch (e) {
  console.error('Plugin error:', e);
}

License

MIT

Credits

Originally based on cordova-plugin-facespoofing by rakaraka029.