So you know by now that the MNIST dataset is comprised of a bunch of handwritten numbers. It is the defacto standard for the Machine Learning community, and it is often used to validate processes. Another group has decided to create another dataset that could be a better replacement. This project is named Fashion-MNIST and is designed to be a simple drop-in replacement. You can get a deeper understanding of the project at https://www.kaggle.com/zalando-research/fashionmnist/data#.
Fashion-MNIST consists of a training set of 60,000 images and labels and a test set of 10,000 images and labels. All images are grayscale and set to 28x28 pixels, and there are 10 classes of images, namely: T-shirt/top, Trouser, Pullover, Dress, Coat, Saldal, Shirt, Sneaker, Bag, and Ankle boot. You can already begin to see that this replacement dataset should work the algorithms...
