# FaceDesk Vision

FaceDesk Vision is a desktop application for real-time facial recognition using a webcam feed. It uses CustomTkinter for the GUI, OpenCV for camera capture and drawing, and face_recognition for deep-learning face embeddings.

## Prerequisites

- Python 3.10 or 3.11 is recommended.
- A working webcam.
- dlib build support, because face_recognition depends on dlib.

### OS notes for dlib

macOS:

```bash
brew install cmake
```

Ubuntu or Debian:

```bash
sudo apt update
sudo apt install -y build-essential cmake python3-dev
```

Windows:

Install Visual Studio Build Tools with the C++ workload, or use a Python distribution where dlib wheels are available for your version.

## Install

```bash
python -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txt
```

On Windows PowerShell:

```powershell
python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install --upgrade pip
pip install -r requirements.txt
```

## Run

```bash
python facedesk_app.py
```

To use a different camera index:

```bash
python facedesk_app.py --camera 1
```

## Storage logic

The app creates these local artifacts automatically:

- `face_registry.sqlite`: SQLite database containing names, 128-dimensional face embeddings, timestamps, and optional reference image paths.
- `registered_faces/`: reference images captured during enrollment.

No cloud service is required. Registered faces persist after restarting the app.

## Enrollment workflow

1. Type a real person name in the enrollment field.
2. Use `Register from camera` to enroll the largest face in the current webcam frame.
3. Use `Register from image` to enroll from a local image file.
4. The app rejects enrollment if no face is found. It warns if multiple faces are found and uses the largest face.

## Recognition workflow

- Camera capture runs continuously in a background thread.
- Recognition runs on skipped frames so the UI remains responsive.
- Known faces are drawn with green boxes and their registered names.
- Unknown faces are drawn with red boxes.
- You can tune recognition tolerance and frame skipping constants near the top of `facedesk_app.py`.

## Privacy

This app stores biometric templates locally. In a production deployment, obtain consent, encrypt the database at rest where required, and publish a clear retention policy.