Face Recognition System

A face recognition system which I implemented for my Diploma di Stato project. After starting the prototype through the attached button, the camera continuously scans the environment for people's faces by using the Viola-Jones algorithm. Following this, the Eigenface method (PCA based) is used for face recognition, which can be trained and then used to successfully identify people.

Roadmap

• Install and configure Raspberry Pi (Jessie)
• Set up SQLite3 database to store face data
• Develop a prototype that continuously scans the environment for faces
• Pre-process face data
• Integrate the Viola-Jones algorithm for face detection
• Implement the Eigenface method for face recognition
• Train the face recognition model
• Test and evaluate with known and unknown faces
• Improve performance and optimisation
  • Optimise the algorithms and code for efficient face data pre-processing and real-time performance on the Raspberry Pi
  • Fine-tune the parameters of the face detection and recognition algorithms to improve accuracy and speed
• Extend the existing implementation to handle multiple faces simultaneously
• Improve user usability (e.g., user management through a button)
• Improve external design
• Translate thesis to English
• Refactor the code

Stack

• Raspberry Pi 3 (Jessie)
• OpenCV
• SQLite3
• Viola-Jones algorithm
• Eigenface method

Project structure

$PROJECT_ROOT
│   # Presentation files (italian only) and other small testing scripts
├── utilities
│   # Custom face data
├── eigenvectors
│   # Dataset
├── training
│   # Key scripts
└── ...

Highlights

Attributions

• Face icons created by Flat Icons - Flaticon

License

Apache 2.0 © Roberto Nacu