Face Detection and Recognition Project

The project is a web application for detecting and recognizing faces in videos. Recognition is implemented using FaceNet, ensuring high accuracy with just one photo of the person of interest.

Contents

1. Project Structure
2. Dependencies
3. Setup and Installation
4. Site Description
5. Features
6. Example Usage
7. Work Visualization
8. License

---

Project Structure

/face_recognition_site
|-- examples/
|   |-- example.png  # Screenshot of the website in action
|   |-- example_video.m4 # Detection Video
|
|-- templates/
|   |-- index.html
|
|-- app.py # Flask app
|
|-- create_embeddings.py # Script for creating embeddings
|
|-- detection.py # Script for creating detection

---

Dependencies

Before running the project, install the following dependencies:

• Python 3.8+
• PyTorch
• OpenCV
• NumPy
• Facenet PyTorch
• Flask
• MoviePy
• Pandas
• Pillow
• Scikit-learn
• Werkzeug

Install the required libraries using the following command:

pip install -r requirements.txt

---

Setup and Installation

1. Clone or download the Sort repository.
2. Clone or download this repository.
3. Install the dependencies:

   pip install -r requirements.txt

4. Set the path to Sort in the path_to_sort variable.

---

Site Description

The website enables face detection and recognition in videos. Users can upload photos of the people they want to recognize. If a matching face is detected in the video, it is labeled with the corresponding name.

---

Features

• Face Detection: Automatically detects all faces in the frame.
• Photo Upload for Recognition: Users can upload photos of people they want to identify.
• Saving Unknown Faces: Faces not recognized among the uploaded photos are saved for later processing.
• Face Identification: Recognized faces are labeled with names in the video.
• Labeling Unknown Faces: Faces seen multiple times are labeled with a unique index.

---

Example Usage

1. The user uploads photos of people. The photo filename should match the person’s name.
2. The website processes the photo using FaceNet and generates a file with embeddings. Uploaded photos are displayed on the site.
3. The user uploads a video.
4. The processed video is displayed on the website with detected faces labeled by name.

---

Work Visualization

---

License

This project is open-source and distributed under the MIT License.