Digit Recognition

Overview

This project focuses on recognizing handwritten digits using a Convolutional Neural Network (CNN) trained on the MNIST dataset. The model predicts digits from 0 to 9 with high accuracy.

Prerequisites

Before you begin, ensure you have met the following requirements:

• Python 3.x
• TensorFlow
• Keras
• NumPy
• Matplotlib

You can install the necessary packages using the following command:

pip install -r requirements.txt

Setup

1. Clone the repository to your local machine:

git clone https://github.com/Gautamhirawat/digitRecognizer.git

2. Navigate to the project directory:

cd digitRecognizer

3. Ensure you have the necessary files: main.py and mnist_cnn_model.h5.

Usage

To run the digit recognition program, execute the following command in your terminal:

python main.py

This will load the pre-trained CNN model from mnist_cnn_model.h5 and start the digit recognition process.

File Descriptions

• main.py: Contains the code for loading the CNN model and recognizing digits from input images.
• mnist_cnn_model.h5: The pre-trained CNN model file.
• testing.py: Contains the code for testing your images . I have given 10 images of digit in this repo.
• digit.ipynb: This file just shows how you can run it and the expected resuts

Results

After running the main.py script, the program will display the prediction and accuracy.

Note

This model is not highly trained , just about 500 images for better results do Train it via notebook or any other python supported Ide .

Contact

If you have any questions or suggestions, please feel free to contact:

• Your Name: Gautam hirawat
• Email: [email protected]

License

This project is licensed under the MIT License. See the LICENSE file for more details.