Advanced Face Recognition System

Advanced Face Recognition System is a sophisticated desktop application developed in Python using the Tkinter library and InsightFace for facial recognition. The system offers both real-time and non-real-time face recognition capabilities, with an intuitive user interface and comprehensive user management features.

Features

Registration

• Register new users using at least 3 face images
• Automatic face embedding calculation and storage
• Support for multiple face angles and expressions

Real-Time Recognition

• Live webcam face detection and recognition
• Real-time identity verification
• Visual feedback with bounding boxes and confidence scores
• Color-coded recognition status (green for known faces, red for unknown)

Non Real-Time Recognition

• Upload and analyze static images
• Support for multiple faces in a single image
• Detailed recognition results with similarity scores
• Visual feedback with highlighted face areas

User Management

• View all registered users
• Delete single or multiple users

Dependencies

To run the project, the following libraries are required:

• Python 3.x
• Tkinter (usually comes with Python)
• OpenCV (cv2)
• NumPy
• PIL (Python Imaging Library)
• InsightFace
• scikit-learn
• threading (built-in)
• pickle (built-in)

You can install the required dependencies using pip:

pip install opencv-python numpy pillow insightface scikit-learn

Installation

1. Clone or download this repository
2. Ensure Python 3.x is installed on your system
3. Install the required dependencies as described above
4. Ensure you have a working webcam for real-time recognition
5. Run the application

How to Use

Run the main file:

python face_recognition_app.py

From the main menu, you can:

Register New User

• Select at least 3 different face images of the user
• Enter the user's name
• Wait for registration confirmation

Real-Time Recognition

• Position face in front of webcam
• View real-time recognition results
• Click "Back to Menu" to exit

Non Real-Time Recognition

• Load an image using "Load Image"
• Click "Verify" to analyze faces
• View recognition results

Manage Registered Users

• View all registered users
• Select users to delete
• Confirm deletions

Technical Details

Face Recognition Process

1. Face Detection

   • Uses InsightFace's buffalo_l model
   • Detects faces in images or video frames
   • Extracts face coordinates

2. Face Embedding

   • Generates 512-dimensional face embeddings
   • Uses cosine similarity for face matching
   • Implements a similarity threshold of 0.6

3. User Registration

   • Stores average embeddings from multiple images
   • Saves data in a pickle file for persistence
   • Validates face presence in registration images

Data Storage

The application stores face embeddings in a local file:

• File name: embeddings.pkl
• Format: Python pickle format
• Content: Dictionary of user names and face embeddings
• Auto-saves after registration/deletion

Screenshots

Menu

Non Real-Time Recognition

Manage Registered Users

Real-Time Recognition (Webcam source)

Contributions

Contributions are welcome! Feel free to:

• Open issues for bugs or suggestions
• Submit pull requests with improvements
• Help with documentation
• Share your experience using the system

License

This project is licensed under the MIT License --->