NIRT ShooterBot

---

Production-ready, modular real-time computer vision framework for robotics and automation.

Requirements

• Python 3.11+
• CUDA-enabled GPU recommended (optional for performance)

Setup and start

1. Create a virtual environment:

   python -m venv .venv

2. Activate it and run the installer:

   .\.venv\Scripts\Activate.ps1
   python -m src.cli.installer

   The installer opens an action menu for Health Check, Install, Wipe and Install, Basic Config, and Advanced Config. It detects NVIDIA hardware and offers CUDA-enabled PyTorch. Select Health Check to inspect the current setup without changing anything.

3. Edit configs/default.yaml as needed.

4. Start the application:

   python -m src --config configs\default.yaml

Features

• YOLO (Ultralytics) primary detector with automatic model download
• Optional face detector (Custom YOLO model)
• Tracking with persistent IDs, velocity, and trajectories
• Generic Arduino serial communication layer (simulation mode)
• Modular architecture: camera, inference, tracking, visualization, config, logging, serial

See docs/Installer.md for setup, dependency, model, and troubleshooting details. Use docs/Scripts.md for the diagnostic Scripts Console and utility-script reference.

License

MIT License