Classifying images in the Fashion-MNIST dataset

This was for an assignment as part of my computer vision module at university.

A convolutional neural network to classify images from the Fashion-MNIST dataset. On average, the network is able to achieve an accuracy of ~92, up from ~89% on the original.

Installation and Usage

This project is best viewed on Google Colab. If this is being run on Colab, you will likely need to change the runtime type. This can be done by going to Runtime > Change runtime type > GPU. Otherwise, the training will take hours!

1. Clone the repository
2. Install required libraries (if not using Colab) - requirements
3. Train the network
4. Look at the results!

Requirements

• tensorflow
• numpy
• matplotlib
• scikitlearn
• jupyter

To install these locally, run the command pip install tensorflow numpy matplotlib scikitlearn jupyter in terminal. This can also be done using conda.

License

This is dual-licensed under MIT and APACHE licenses.

Credit

Credits to @radenjezic153 for building the basic framework which I was able to improve upon.