🔢 Handwritten Digit Classifier

This is a beginner-friendly project where I built a simple digit recognition app using a neural network trained on the MNIST dataset. The goal was to understand the end-to-end workflow of a Machine Learning project — from model creation to deployment in a simple graphical interface.

🧠 This was my first Machine Learning project!
You’ll notice I included lots of comments in the code to help myself understand each step.
The purpose was educational — to learn deeply, not just make it work. 😊

---

💡 How it works

• You draw a number (0–9) in a canvas with your mouse.
• The image is preprocessed and passed to a trained PyTorch model.
• The model predicts which number it thinks you drew.
• The result is displayed instantly!

---

🧠 Model

The model is a simple feedforward neural network implemented with PyTorch:

• Input layer: 28×28 = 784 pixels (flattened)
• Hidden layer: 15 neurons, ReLU activation
• Output layer: 10 neurons (digits 0–9), with raw scores (logits)

class NeuralNetwork(nn.Module):
    def __init__(self):
        super().__init__()
        self.flatten = nn.Flatten()
        self.network = nn.Sequential(
            nn.Linear(28*28, 15),
            nn.ReLU(),
            nn.Linear(15, 10)
        )

    def forward(self, x):
        x = self.flatten(x)
        return self.network(x)

The model was trained on the MNIST dataset.

---

✨ GUI App

• Built with Tkinter (Python's standard GUI library).
• You can draw directly on a canvas.
• The app converts your drawing into a 28×28 grayscale image, normalizes it, and passes it to the model.

---

📂 Files

• model.py: The neural network definition.
• training.py: Script to train the model.
• app.py: The graphical user interface.
• model.pth: Saved PyTorch model (trained weights).

---

⚙️ Requirements

Install dependencies with:

pip install -r requirements.txt

tkinter comes pre-installed with Python (especially on Windows). No need to install it with pip.

---

🚀 How to Run

1. Train the model (or use the included model.pth):

   python training.py

2. Run the app:

   python app.py

---

📸 Preview

---

📝 License

This project is licensed under the MIT License – see the LICENSE file for details.

---

💬 Final note

Thank you for checking out my project!
This was an exciting start in my Machine Learning journey — feel free to reach out if you have suggestions or feedback 😊💖