face_detection_tflite

A pure Dart/Flutter implementation of Google's MediaPipe face detection and facial landmark models using TensorFlow Lite. This package provides on-device face and landmark detection with minimal dependencies, just TensorFlow Lite and image.

Bounding Box Example:

Mesh (468-Point) Example:

Landmark Example:

Iris Example:

Table of Contents

• Features
• Quick Start
• Models
• Types
• Example
• Inspiration

Features

• On-device face detection (multiple SSD variants)
• 468-point face mesh, face landmarks, iris landmarks and bounding boxes
• All coordinates are in absolute pixel coordinates (Point<double>)
  • x ranges from 0 to image.width
  • y ranges from 0 to image.height
  • Ready to use co-ordinates without any scaling
• Truly cross-platform: compatible with Android, iOS, macOS, Windows, and Linux
• The example/ app illustrates how to detect and render results on images: bounding boxes, a 468-point face mesh, and iris landmarks.

Quick Start

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

Future main() async {
  // 1. initialize
  final detector = FaceDetector();
  await detector.initialize(model: FaceDetectionModel.backCamera);

  // 2. detect
  final imageBytes = await File('path/to/image.jpg').readAsBytes();
  final result = await detector.detectFaces(imageBytes);

  // 3. access results
  for (final face in result.faces) {
    final landmarks = face.landmarks;
    final bbox = face.bboxCorners;  
    final mesh = face.mesh;     
    final irises = face.irises;
    
    final leftEye = landmarks[FaceIndex.leftEye];
    final rightEye = landmarks[FaceIndex.rightEye];

    print('Left eye: (${leftEye.x}, ${leftEye.y})');
  }

  // 4. clean-up
  detector.dispose();
}

Face Detection Modes

This app supports three detection modes that determine which facial features are detected:

Mode	Features	Est. Time per Face*
Full (default)	Bounding boxes, landmarks, 468-point mesh, iris tracking	~80-120ms
Standard	Bounding boxes, landmarks, 468-point mesh	~60ms
Fast	Bounding boxes, landmarks	~30ms

*Est. times per faces are based on 640x480 resolution on modern hardware. Performance scales with image size and number of faces.

Code Examples

// full mode (default): bounding boxes, 6 basic landmarks + mesh + iris
// note: full mode provides superior accuracy for left and right eye landmarks
// compared to fast/standard modes. use full mode when precise eye landmark
// detection is required for your application. trade-off: longer inference
await _faceDetector.detectFaces(bytes);

// standard mode: bounding boxes, 6 basic landmarks + mesh. inference time 
// is faster than full mode, but slower than fast mode.
await _faceDetector.detectFaces(bytes, mode: FaceDetectionMode.standard);

// fast mode: bounding boxes + 6 basic landmarks only. fastest inference 
// time of the three modes.
await _faceDetector.detectFaces(bytes, mode: FaceDetectionMode.fast);

Try the sample code from the pub.dev example tab to easily compare modes and inferences timing.

Models

You can choose from several detection models depending on your use case:

• FaceDetectionModel.backCamera: larger model for group shots or images with smaller faces (default).
• FaceDetectionModel.frontCamera: optimized for selfies and close-up portraits.
• FaceDetectionModel.short: best for short-range images (faces within ~2m).
• FaceDetectionModel.full: best for mid-range images (faces within ~5m).
• FaceDetectionModel.fullSparse: same detection quality as full but runs up to 30% faster on CPU (slightly higher precision, slightly lower recall).

Types

• FaceResult contains bboxCorners, mesh, irises and landmarks.
• face.landmarks is a Map<FaceIndex, Point<double>>, where FaceIndex is one of:
  • FaceIndex.leftEye
  • FaceIndex.rightEye
  • FaceIndex.noseTip
  • FaceIndex.mouth
  • FaceIndex.leftEyeTragion
  • FaceIndex.rightEyeTragion
• All coordinates are absolute pixel positions, ready to use for drawing or measurement.

Example

The sample code from the pub.dev example tab includes a Flutter app that paints detections onto an image: bounding boxes, landmarks, mesh, and iris. The example code provides inference time, and demonstrates when to use FaceDetectionMode.standard or FaceDetectionMode.fast.

Inspiration

At the time of development, there was no open-source solution for cross-platform, on-device face and landmark detection. This package took inspiration and was ported from the original Python project patlevin/face-detection-tflite. Many thanks to the original author.