A professional desktop application for extracting telemetry data from racing/motorsport onboard videos using advanced OCR and computer vision.

• 🚀 Advanced OCR Engine - Uses EasyOCR for best accuracy, will test other engines in the future
• 📊 Multiple Metrics - Extract speed, G-force, per-wheel torque(beta), and more
• 🎯 Interactive ROI Selection - Draw regions of interest directly on the video
• ⚡ Real-time Preview - See extracted values overlaid on video playback
• 🎨 Modern UI - Polished dark theme with a nice styling
• 📦 Standalone Executable - Distributable to others without requiring Python

Coming soon

1. Download the latest release from the Releases page
2. Run FB_Data_Scraper_vX.X.X.exe (version number will be in the filename)
3. No Python installation required!

💡 Note: The executable is automatically built and released on every commit to master.

To enable GPU acceleration:

1. Have an NVIDIA GPU (GTX 10 series or newer recommended)
2. Install CUDA Toolkit 11.8
3. That's it! The exe will detect CUDA and use your GPU automatically

The exe works without CUDA - it's just slower. No extra setup needed for CPU-only usage.

# Clone the repository
git clone https://github.com/yourusername/fb_data_scraper.git
cd fb_data_scraper

# Create virtual environment
python -m venv venv
venv\Scripts\activate  # Windows
source venv/bin/activate  # Linux/Mac

# Install dependencies
pip install -r requirements.txt

For GPU acceleration when running from source:

# For EasyOCR GPU support:
pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118

python run.py

1. Open a Video - File → Open Video (or Ctrl+O)
2. Initialize OCR - Tools → Initialize OCR (first time only, will auto-download models)
3. Select ROIs - Click "Select ROI" for each metric and draw a box around the data area
4. Play Video - Use transport controls or Space to play/pause
5. Export Data - File → Export Data to save extracted values to CSV

To create a distributable executable:

# Install PyInstaller
pip install pyinstaller

# Build the executable
python build.py

# Or with options
python build.py --clean    # Clean first
python build.py --onedir   # Build as folder (easier to debug)

The executable will be in the dist/ folder.

fb_data_scraper/
├── run.py                  # Application entry point
├── build.py                # Build script for executable
├── requirements.txt        # Python dependencies
├── fsae_extractor.spec     # PyInstaller configuration
├── src/
│   ├── core/
│   │   ├── ocr_engine.py   # PaddleOCR/EasyOCR wrapper
│   │   ├── preprocessor.py # Image preprocessing pipeline
│   │   ├── video.py        # Video handling
│   │   └── extractors/     # Metric extractors
│   │       ├── base.py     # Base extractor class
│   │       ├── speed.py    # Speed OCR extractor
│   │       ├── gforce.py   # G-force OCR extractor
│   │       └── torque.py   # Torque color analyzer
│   ├── gui/
│   │   ├── main_window.py  # Main application window
│   │   ├── styles.py       # Modern UI styling
│   │   ├── splash.py       # Splash screen
│   │   └── widgets/        # Custom UI widgets
│   └── config/
│       └── settings.py     # Configuration management
├── assets/                 # App icons and images
└── dist/                   # Built executables (after build)

The app includes optimized presets for different video types:

• EasyOCR - Simpler installation, used as fallback
• PaddleOCR - legacy versions may still have Paddle installed, but it isn't currently used due to innaccurate readings

MIT License - see LICENSE file for details.