FaceMatch

FaceMatch is an AI-powered Django application that uses the DeepFace library to find facial similarities by comparing image embeddings. The project offers two endpoints: one for saving user images as embeddings and another for comparing a given image against saved embeddings to find a matching face. This tool is ideal for applications such as identity verification, duplicate detection, or face matching in various organizational contexts.

Features

• Facial Embedding Generation: Converts uploaded images into numerical embeddings using deep learning models.
• Face Similarity Comparison: Compares new images with stored embeddings to find similar faces, providing a similarity score.
• Multiple Model Support: Leverages various models like Facenet, VGG-Face, and others from the DeepFace library.
• Organizational Segmentation: Manages embeddings on a per-organization basis for better data isolation.

Tech Stack

• Backend: Django, Django REST Framework
• Machine Learning: DeepFace for facial recognition
• Database: SQLite (default) or any other supported by Django ORM

Installation

1. Clone the repository:

   git clone https://github.com/your-username/FaceMatch.git
   cd FaceMatch

2. Set up a virtual environment:

   python -m venv venv
   source venv/bin/activate # On Windows use `venv\Scripts\activate`

3. Install dependencies:

   pip install -r requirements.txt

4. Apply migrations:

   python manage.py migrate

5. Run the server:

   python manage.py runserver

API Endpoints

1. Embedding Endpoint

• URL: /api/embeddings/
• Method: POST
• Description: Saves images for a specific organization and generates embeddings.
• Request:

  {
    "organization": "string",
    "images": [file]
  }

• Response:

  {
    "embed_id": "integer",
    "organization": "string",
    "embedding": "string",
    "created_at": "datetime"
  }

2. Face Comparison Endpoint

• URL: /api/face-compare/
• Method: POST
• Description: Compares an uploaded image with the saved embeddings for a specific organization.
• Request:

  {
    "organization": "string",
    "image": file
  }

• Response:

  {
    "embed_id": "integer",
    "similarity_score": "float"
  }

Project Structure

• views.py: Contains API views for embedding generation and face comparison.
• services.py: Handles core logic for embedding generation, comparison, and utility functions for processing images.
• utils.py: Provides helper functions for saving images, embedding processing, and finding similar faces.
• models.py: Defines database models for storing embeddings and associated image metadata.
• serializers.py: Serializes input data for the API endpoints.

Example Usage

Save Embeddings

To save embeddings for an organization, send a POST request to /api/embeddings/ with the following body:

curl -X POST http://localhost:8000/api/embeddings/ \
-F "organization=MyOrganization" \
-F "images=@/path/to/image1.jpg" \
-F "images=@/path/to/image2.jpg"

Compare Faces

To compare a face, send a POST request to /api/face-compare/:

curl -X POST http://localhost:8000/api/face-compare/ \
-F "organization=MyOrganization" \
-F "image=@/path/to/query_image.jpg"

Contributing

Feel free to open issues or submit pull requests for improvements. Contributions to enhance functionality or fix bugs are always welcome.