Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms

TL;DR

Fashion-MNIST is a new, standardized dataset of 70,000 fashion images designed to replace MNIST for benchmarking machine learning algorithms, facilitating consistent evaluation across research.

Contribution

It introduces a new dataset with fashion images that mimics MNIST's structure, enabling direct comparison and benchmarking of machine learning models on fashion classification tasks.

Findings

Provides a challenging benchmark for ML algorithms.

Facilitates consistent evaluation across models.

Enables research in fashion image recognition.

Abstract

We present Fashion-MNIST, a new dataset comprising of 28x28 grayscale images of 70,000 fashion products from 10 categories, with 7,000 images per category. The training set has 60,000 images and the test set has 10,000 images. Fashion-MNIST is intended to serve as a direct drop-in replacement for the original MNIST dataset for benchmarking machine learning algorithms, as it shares the same image size, data format and the structure of training and testing splits. The dataset is freely available at https://github.com/zalandoresearch/fashion-mnist