Handwritten Digits Recognition

A simple deep learning project using TensorFlow and Keras to classify handwritten digits (0–9) from the MNIST dataset. This project demonstrates how to build, train, and evaluate a basic neural network for image classification.

Model Architecture

• Input Layer: 28x28 pixels (flattened to 784)
• Hidden Layer: Dense(128, ReLU)
• Output Layer: Dense(10, Softmax)

---

Dataset

• Training images: 60,000
• Test images: 10,000
• Image size: 28x28 (grayscale)

---

How to Run

1. Clone this repo:

   git clone https://github.com/AdarshZolekar/Handwritten-Digits-Recognition.git
   cd Handwritten-Digits-Recognition

2. Install dependencies:

   pip install -r Requirements.txt
   

3. Run the project:

   python MNIST-Digits-Recognition.py
   

---

Requirements

• Python 3.x

• TensorFlow

• NumPy

• Matplotlib

Install them with:

pip install tensorflow numpy matplotlib

---

Learning Outcome

• Basics of neural networks

• Hands-on with TensorFlow & Keras

• Experience in image preprocessing.

---

Future Improvements

• Switch from a simple dense network to a Convolutional Neural Network (CNN) for higher accuracy

• Add a Streamlit/Gradio app for interactive digit drawing and real-time predictions

• Deploy model on web/mobile using TensorFlow Lite or Flask API

• Experiment with additional datasets (letters, symbols) for broader recognition

• Hyperparameter tuning and regularization to reduce overfitting.

---

License

This project is open-source and free to use for educational purposes.

---

Contributions

Feel free to open issues or submit pull requests for improvements or bug fixes!