This project is a handwritten digits classifier built using PyTorch. It leverages the MNIST dataset and demonstrates a machine learning pipeline from data loading and preprocessing to model training and evaluation. The project also includes a user-friendly web interface built with Streamlit, allowing users to interactively test the model with their own handwritten digit inputs.

The classifier is built using a Convolutional Neural Network (CNN), a common architecture for image classification tasks. The model is designed to recognize digits from 0 to 9 based on the MNIST dataset.

• Architecture: The CNN model consists of several convolutional layers followed by fully connected layers.
• Framework: PyTorch is used for model implementation and training.

The model is trained and evaluated on the MNIST dataset, which consists of:

• Training Set: 60,000 images of handwritten digits.
• Test Set: 10,000 images of handwritten digits.

The images are grayscale and of size 28x28 pixels.

The model achieves high accuracy on the MNIST test set. Performance metrics and accuracy details are logged during training and evaluation phases.

You can access the Streamlit app here: dubeyayushi-mnist-handwritten-digits-classifie-mnist-app-plssff.streamlit.app

The Streamlit application provides an interactive interface for users to test the classifier. Users can:

• Upload an Image: Load a handwritten digit image.
• Predict Digit: The model will predict the digit and display the result.

• Image Upload: Drag and drop or browse to select an image file.
• Prediction: Real-time classification and display of results.

Ensure you have Python 3.8 or higher installed.

1. Clone the Repository:

   git clone https://github.com/your-username/handwritten-digits-classifier.git
   cd handwritten-digits-classifier
   

2. Create a Virtual Environment (optional but recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`
   

3. Install Dependencies:

   Make sure your requirements.txt does not specify strict versions if you want pip to handle compatibility automatically.

   pip install -r requirements.txt
   

4. Run the Streamlit App:

   streamlit run mnist_app.py
   

1. Open the Streamlit App: Follow the instructions above to start the app.

2. Upload a Handwritten Digit Image: Use the upload button to select an image file from your computer.

3. View Prediction: The model will predict the digit and display the result on the screen.

Contributions are welcome! If you have suggestions, improvements, or fixes, please feel free to submit a pull request or open an issue.

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

• MNIST Dataset: A classic dataset used for training image processing systems.
• PyTorch: For providing the deep learning framework.
• Streamlit: For enabling easy web app creation and deployment.

This project was completed as part of the advanced Machine Learning Nanodegree Program by Udacity and AWS.