🎯 Pitch AI

A JavaScript web application that allows you to play the classic card game Pitch against AI opponents powered by neural networks trained with reinforcement learning.

🚀 Features

• Full Pitch Game Implementation: Complete 4-player Pitch game with bidding and playing phases
• AI Opponents: Three AI players using trained neural networks
• Interactive UI: Clean, modern interface with card visualization
• Real-time Gameplay: Smooth turn-based gameplay with visual feedback
• Auto-play Mode: Watch AI vs AI games or play manually
• Responsive Design: Works on desktop and mobile devices

🏗️ Architecture

The application consists of:

• Frontend: HTML5, CSS3, and vanilla JavaScript
• AI Models: TensorFlow.js models for bidding and playing decisions
• Game Logic: Complete Pitch game implementation
• State Management: Real-time game state tracking

🛠️ Setup Instructions

1. Clone or Download the Project

git clone <repository-url>
cd pitch

2. Run the Application

Simply open index.html in your web browser.

Note: For best performance and to avoid CORS issues, consider using a local server:

# Python 3
python -m http.server 8000

# Python 2
python -m SimpleHTTPServer 8000

# Node.js (if you have it installed)
npx http-server

Then open http://localhost:8000 in your browser.

🎮 How to Play

Game Rules

Pitch is a trick-taking card game where:

• 4 players play in teams of 2
• Each player is dealt 6 cards
• Players bid on how many points they can win
• Trump suit is determined by the first card played
• Points are awarded for winning tricks and special cards

Controls

• New Game: Start a new game
• Auto Play: Let the AI play automatically
• Reset: Reset scores and return to waiting state
• Bidding: Click bid buttons during bidding phase
• Playing: Click cards in your hand to play them

Game Flow

1. Bidding Phase: Players bid on points (Pass, 2, 3, or 4)
2. Playing Phase: Winner leads first trick, others follow suit
3. Scoring: Points awarded based on tricks won and bid success

🔧 Customization

Adding to Existing Webpages

The Pitch AI demo can be easily integrated into existing HTML pages:

<!-- Include TensorFlow.js -->
<script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@4.15.0/dist/tf.min.js"></script>

<!-- Include the game -->
<script src="pitch-ai.js"></script>

<!-- Add the game container -->
<div id="pitch-game-container">
    <!-- Copy the game-container div from index.html -->
</div>

<!-- Initialize the game -->
<script>
    const pitchGame = new PitchAI();
</script>

Styling Customization

Modify styles.css to match your website's theme:

• Color schemes
• Typography
• Layout dimensions
• Card designs

📁 Project Structure

pitch/
├── index.html              # Main HTML file
├── styles.css              # CSS styling
├── pitch-ai.js            # Main game logic
├── models/                 # TensorFlow.js models
│   ├── bidding_model/     # Bidding AI model
│   └── playing_model/     # Playing AI model
├── .gitignore             # Git ignore rules
└── README.md              # This file

🐛 Troubleshooting

Models Not Loading

• Check browser console for errors
• Ensure models directory exists with proper structure
• Verify TensorFlow.js is loaded correctly

Game Not Starting

• Check JavaScript console for errors
• Ensure all files are in the same directory
• Try using a local server instead of file:// protocol

Performance Issues

• Use a modern browser (Chrome, Firefox, Safari, Edge)
• Ensure sufficient RAM for model loading
• Consider model quantization for better performance

🔮 Future Enhancements

• Model Training: Web-based model training interface
• Multiplayer: Online multiplayer support
• Tournaments: AI tournament mode
• Analysis: Game replay and analysis tools
• Customization: More game variants and rules

📚 Technical Details

Neural Network Architecture

• Bidding Model: 56 inputs → 128 hidden → 4 outputs
• Playing Model: 184 inputs → 256 hidden → 52 outputs
• Activation: LeakyReLU for hidden layers, Softmax for outputs

State Representation

• Bidding State: 52-card hand + 4-position encoding
• Playing State: Hand + trump + current trick + history + bidding info

AI Decision Making

• Policy gradient with actor-critic architecture
• Legal move masking for valid actions
• Stochastic sampling for exploration

🤝 Contributing

Contributions are welcome! Areas for improvement:

• Bug fixes and performance optimizations
• UI/UX enhancements
• Additional game features
• Model improvements
• Documentation updates

📄 License

This project is open source. Free to use, modify, and distribute as needed.

🙏 Acknowledgments

• Original Pitch game rules and mechanics
• TensorFlow.js team for browser-based ML
• Card game enthusiasts for inspiration

---

Enjoy playing Pitch against AI! 🎯♠️♥️♦️♣️