DigitMind

DigitMind 🧠 is a Python-based web app that recognizes handwritten digits in real time. Users can draw a digit on the screen, and an AI model will attempt to guess which number it is. This project uses a fully connected neural network trained on the MNIST dataset, resized to 64x64 for flexibility.

---

Features

• ✏️ Draw digits directly in the browser
• 🤖 Neural network powered by PyTorch
• ⚖️ Customizable input size (64x64 pixels)
• ✨ Clean Streamlit UI
• ✅ Easy to extend and retrain

---

Installation

1. Clone the repository:

   git clone https://github.com/your-username/DigitMind.git
   cd DigitMind

2. (Optional) Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # Windows: .\venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Train the model:

   python train.py

5. Run the app:

   python -m streamlit run app.py

---

Requirements

• Python 3.8+
• PyTorch
• torchvision
• Streamlit
• streamlit-drawable-canvas
• OpenCV
• Pillow

---

File Structure

DigitMind/
├── app.py               # Streamlit app
├── train.py             # Training script
├── model.py             # Neural network model definition
├── digit_model.pth      # Saved trained model (after running train.py)
├── requirements.txt     # Dependencies
├── debug_digit.png      # Optional debug image
└── README.md            # You're here :)

---

Author

Created by SynvexAI. Contributions and feedback welcome!