Recognition of the MNIST dataset

Digits (mnist datasets) and fashion items (fashion_mnist datasets) recognition using python based KNN, neural network(NN), and convolutional neural network(CNN) algorithms

Running Guide

This project is based on the Python programming language and primarily utilizes standard libraries like Tensorflow, PIL, numpy, scikit-learn and os

Environment Setup

Download the requirements.txt and install the required Python libraries. Please note all my 4 projects share the same requirements.txt. If you have done the installation for one project, you can skip it for the other 3 projects

# install all packages using requirements.txt
python -m pip install -r requirements.txt

Training the Model and check the accuracy on testing datasets

• If you want to train your model for MNIST datasets, you can run KNN_mnist.py, neural_network_mnist.py, or CNN_mnist.py in the folder. The model will train itself using the 60000 trainning images and then test itself using the 10000 testing images in MNIST datasets. I have selected a set of resonable model parameters for all the models in the codes but you can always try yourselves with a different set of parameters. For KNN_mnist.py, you can amend the line 25. For neural_network_mnist.py, you can amend the line 25.
• If you want to train your model for Fashion_MNIST datasets, you can run KNN_fashion_mnist.py, neural_network_fashion_mnist.py, or CNN_fashion_mnist.py in the folder. The model will train itself using the 60000 trainning images and then test itself using the 10000 testing images in Fashion_MNIST datasets. I have selected a set of resonable model parameters for all the models in the codes but you can always try yourselves with a different set of parameters.