A Laravel web application for browsing and voting on video files. Videos are displayed as animated GIF previews that users can click to watch the full video. Includes offline face recognition to filter videos by the people in them.

• PHP 8.0+
• Composer
• MySQL
• Node.js & npm
• FFmpeg (must be in PATH)
• Python 3.8+ (for face recognition)

# Install PHP dependencies
composer install --optimize-autoloader --no-dev

# Install Node dependencies
npm install

# Copy environment file and configure
cp .env.example .env
php artisan key:generate

# Run migrations
php artisan migrate --force

# Link storage
php artisan storage:link

# Symlink video directory (adjust path to your video source)
ln -s video_base_dir public/storage/video

# Set permissions (Linux)
chgrp -R www-data storage bootstrap/cache
chmod -R ug+rwx storage bootstrap/cache

Configure your .env file with MySQL credentials:

DB_CONNECTION=mysql
DB_HOST=127.0.0.1
DB_PORT=3306
DB_DATABASE=video
DB_USERNAME=video
DB_PASSWORD=video

Ensure the following directories exist inside public/storage/:

public/storage/
├── video/      # Source video files (.mp4)
├── gif/        # Generated GIF previews
├── css/        # Stylesheets
└── faces/      # Auto-generated face thumbnails

The gif.sh script creates animated GIF previews from videos. Run it from public/storage/:

cd public/storage
bash gif.sh

For each video it extracts three 20-second clips (beginning, middle, end), combines them, and generates a sped-up GIF.

php artisan serve

Visit http://localhost:8000. All pages require login.

* * * * * cd /path-to-your-project && php artisan schedule:run >> /dev/null 2>&1

Or locally:

php artisan schedule:work

Each video has a vote counter. Click the submit button on the video player page to add +1 vote.

Offline face recognition powered by Python's face_recognition library (uses dlib). No data is sent to any external service — everything runs locally.

1. Extracts frames from each video using FFmpeg
2. Detects faces in each frame using dlib's HOG-based detector
3. Encodes each face as a 128-dimensional vector
4. Compares against known faces — matches existing people or creates new ones
5. Saves relationships to the database and generates face thumbnails
6. All views show clickable person buttons to filter videos

pip install cmake
pip install dlib
pip install face_recognition numpy mysql-connector-python Pillow

Note: dlib requires CMake and a C++ compiler. On Windows, install Visual Studio Build Tools with the "Desktop development with C++" workload.

Via artisan:

php artisan faces:scan

Options:

Example with custom settings:

php artisan faces:scan --tolerance=0.5 --frames=8

Or run the Python script directly:

python scripts/scan_faces.py --video-dir public/storage/video --tolerance 0.6 --frames 5

Detected people are unnamed by default (shown as "Person #1", etc.). Click a person button on any page, then use the rename form on the person page.

The scanner skips already-processed videos. To re-scan everything:

TRUNCATE file_person;

Then run php artisan faces:scan again.

If you run into issues:

php artisan optimize

To disable public registration, comment out in config/fortify.php:

//Features::registration()

• Backend: Laravel 8, Fortify, Jetstream, Sanctum
• Frontend: Blade templates, Bootstrap 5, Inertia.js + Vue 3 (auth pages)
• Database: MySQL
• Media: FFmpeg, freezeframe.js
• Face Recognition: Python, dlib, face_recognition (fully offline)