https://github.com/thatlinuxguyyouknow/mnist-digit-classifier

A simple KERAS based CNN trained on the MSINT Dataset
https://github.com/thatlinuxguyyouknow/mnist-digit-classifier

Last synced: over 1 year ago
JSON representation

A simple KERAS based CNN trained on the MSINT Dataset

• Host: GitHub
• URL: https://github.com/thatlinuxguyyouknow/mnist-digit-classifier
• Owner: ThatLinuxGuyYouKnow
• Created: 2025-03-06T10:59:41.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2025-03-09T12:49:50.000Z (over 1 year ago)
• Last Synced: 2025-03-09T13:34:23.868Z (over 1 year ago)
• Language: Python
• Size: 7.19 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# MNIST Digit Classifier

[![Python](https://img.shields.io/badge/python-3.x-blue.svg)](https://www.python.org/)

[![TensorFlow](https://img.shields.io/badge/TensorFlow-2.x-orange.svg)](https://www.tensorflow.org/)

[![NumPy](https://img.shields.io/badge/NumPy-1.x-green.svg)](https://numpy.org/)

[![Pillow](https://img.shields.io/badge/Pillow-x.x-purple.svg)](https://pillow.readthedocs.io/en/stable/)

This project is a backend-only Python application that trains, evaluates, and tests a Convolutional Neural Network (CNN) model for classifying handwritten digits from the MNIST dataset.  It uses TensorFlow/Keras for model building and training, NumPy for numerical operations, and Pillow for image manipulation.

## Accessible Routes & Methods

The application is run from the command line and doesn't have traditional web routes.  Instead, it offers the following command-line actions:

* **`python main.py train`**:  Trains the MNIST model using the MNIST dataset.  No parameters are required. This command saves the model to `mnist_model.h5`

* **`python main.py evaluate`**: Loads the model from `mnist_model.h5` and evaluates its performance on the MNIST test dataset. Prints the test accuracy. No parameters are required.

* **`python main.py test`**: Loads the model from `mnist_model.h5` and uses it to predict the digit in `test.png`.  Requires a `test.png` image in the same directory.  The image should be a 28x28 grayscale image of a handwritten digit.

* **`python main.py easy`**: Loads the model from `mnist_model.h5` and uses it to predict the digit in `test_easy.png`. Requires a `test_easy.png` image in the same directory.  The image should be a 28x28 grayscale image of a handwritten digit.

All actions use the `GET` method implicitly through command-line arguments.  No parameters are passed directly as part of a request, but the presence or absence of specific files (e.g., `test.png`) indirectly influences the behavior.  Error handling is implemented to gracefully handle the absence of a trained model or required image files.