Handwritten-Digit-Classifier

This is a simple handwritten digit classifier which uses Convolutional Neural Network to classify the input image and predict the digit present in the image.

---

This project implements a simple Convolutional Neural Network (CNN) using TensorFlow and Keras to classify handwritten digits from the MNIST dataset. The trained model is saved as mark1.keras.

Features

• Uses a CNN model for handwritten digit recognition.
• Trained on the MNIST dataset with 10 epochs.
• Model is saved as mark1.keras for inference.
• Includes a script to classify a given handwritten digit image.

Installation

1. Clone the repository:

   git clone https://github.com/your-username/handwritten-digit-classification.git
   cd handwritten-digit-classification

2. Install dependencies:

   pip install tensorflow numpy matplotlib pandas scikit-learn

Training the Model

To train the model, run:

python digit_classification.py

This will train the CNN on the MNIST dataset and save the model as mark1.keras.

Using the Model for Prediction

To classify a handwritten digit image, run:

python digit_class.py

Then enter the path to the image when prompted.

Model Architecture

• 3 Convolutional Layers (with ReLU activation and max pooling)
• Flatten Layer
• Fully Connected Dense Layers
• Softmax Output Layer (for 10-class classification)

Example Usage

If you have an image digit.png containing a handwritten digit, you can classify it as follows:

python digit_class.py

Then, enter:

path/to/digit.png

The script will output the predicted digit.

License

This project is licensed under the MIT License.

---