🃏 Playing Card Recognition

A deep learning project for playing card recognition using Convolutional Neural Networks (CNNs) in PyTorch. Includes scripts for training, evaluation, testing, and real-time camera-based recognition.

---

📁 Project Structure

├── cameraRecognition.py       # Real-time card recognition via webcam
├── modelTesting.ipynb         # Model evaluation and visualization notebook
├── neuralNetwork.py           # Neural network architecture, training, and utilities
├── transform.py               # Data augmentation and transformation helpers
├── acc81.5/                   # Trained models, results, plots, and confusion matrices
├── dataset/                   # Dataset (train/valid/test splits)
└── README.md                  # Project documentation

🗂️ Dataset Structure

dataset/
├── train/
├── valid/
└── test/

Each subfolder should contain one folder per card class with images inside.

---

🧠 Model Training

Train your neural network using the functions in NeuralNetworkMain.py.

---

📊 Model Evaluation

Use the modelTesting.ipynb notebook to:

• Load the best model (acc81.5/best_model.pth)
• Evaluate accuracy on the test set
• Generate and visualize confusion matrices for:
  • Color
  • Rank
  • Suit
  • Full classes

---

🎥 Real-Time Recognition

Run the following command:

python cameraRecognition.py

The script will activate your webcam and start recognizing visible cards in real time.

---

🖼️ Results (Examples)

🔍 Live Recognition Sample:

📈 Learning Curve:

♠️♥️♦️♣️ Confusion Matrices – Rank / Suit:

---

📦 Contents of acc81.5/

• best_model.pth — best performing trained model
• Training/validation loss and accuracy plots
• Confusion matrices for various categories

---

📄 License

This project is intended for educational and research purposes only.