Where is my face ?

A Flutter application that uses AI-powered face detection to identify and extract faces from camera feed or images. Built with the YuNet model for accurate real-time face detection.

Features

• ✅ Real-time Face Detection - Detect multiple faces simultaneously using camera feed
• ✅ Face Confidence Score - Display confidence percentage for each detected face
• ✅ Face Extraction - Extract and save detected faces as separate images
• ✅ Multiple Camera Support - Switch between front and back cameras
• ✅ Flash Control - Toggle flash on/off for better lighting
• ✅ Modern UI - Clean, dark-themed interface with intuitive controls
• ✅ Batch Processing - Select and save multiple faces at once
• ✅ Detection Modes - Switch between Video, Detect, and Portrait modes

Screenshots

🛠️ Tech Stack

• Flutter SDK: ^3.9.2
• Camera: ^0.11.3 - Camera functionality
• OpenCV Dart: ^1.4.3 - Image processing and face detection
• YuNet Model: ONNX-based face detection model

Prerequisites

Before running this project, ensure you have:

• Flutter SDK installed (3.9.2 or higher)
• Android Studio / Xcode (for mobile development)
• Camera permissions configured

Getting Started

Installation

1. Clone the repository

   git clone <repository-url>
   cd face_extractor

2. Install dependencies

   flutter pub get

3. Run the app

   flutter run

Configuration

The YuNet model file (yunet.onnx) is located in the assets/ directory and is automatically loaded when the app starts.

Project Structure

lib/
├── main.dart                              # App entry point
├── components/
│   └── extracted_faces_dialog.dart       # Dialog for displaying extracted faces
├── models/
│   └── face_detection_result.dart        # Face detection data model
├── painter/
│   └── face_detection_painter.dart       # Custom painter for face overlays
├── screens/
│   └── camera_screen.dart                # Main camera and detection screen
└── services/
    └── face_detector_service.dart        # Face detection service using YuNet

assets/
└── yunet.onnx                            # YuNet face detection model

Features Breakdown

Camera Screen

• Top Bar: Settings, app title, and flash toggle
• Detection Overlay: Real-time bounding boxes with confidence scores
• Status Indicator: Shows number of detected faces
• Mode Selector: Switch between VIDEO, DETECT, and PORTRAIT modes
• Camera Controls: Gallery access, capture button, and camera flip

Face Extraction Dialog

• Single Face Mode: Shows extracted face with save and retake options
• Multiple Faces Mode: Grid of selectable faces with batch save functionality
• Smart Selection: Tap to select/deselect individual faces
• Action Buttons: Save selected faces, save all faces, or retake picture

To-Do List

High Priority

• Save to Device Storage - Implement actual file saving functionality with proper permissions
• Image Gallery View - Create a screen to view all previously extracted faces

Low Priority

• Settings Screen - Add customizable detection sensitivity and other preferences
• Dark/Light Theme - Add theme switching capability
• Share Functionality - Share extracted faces directly to other apps
• Video Recording - Record video while detecting faces

Completed

• Real-time face detection
• Face extraction functionality
• Camera switching (front/back)
• Flash control
• Multiple face detection
• Confidence score display
• Modern UI implementation
• Face selection in dialog
• Pick Image from Gallery

Configuration

Android Permissions

Add the following permissions to android/app/src/main/AndroidManifest.xml:

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

🎯 How It Works

1. Camera Initialization: The app initializes the camera and loads the YuNet model
2. Frame Processing: Every 200ms, a frame is captured and processed
3. Face Detection: YuNet model detects faces and returns bounding boxes with confidence scores
4. Visualization: Detected faces are highlighted with cyan boxes and confidence labels
5. Extraction: On capture, faces are cropped from the image using OpenCV
6. Display: Extracted faces are shown in a dialog with save options

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is open source and available under the MIT License.

Acknowledgments

• YuNet Model - Face detection model by OpenCV
• OpenCV Dart - Dart bindings for OpenCV
• Flutter Team - For the amazing framework

Contact

For questions or support, please open an issue in the repository.

---

Version: 1.2.0
AI: Active
Status: In Development