Extend Fashion MNIST yourself — Machine Learning DYI

How to utilize Bing Search for ML.

By getting to this article you probably know what MNIST and Fashion MNIST are and how those datasets can kickstart your Machine Learning (ML) projects.
By the end of this article you’ll know how with Azure and little bit of Python code you can build your own training dataset for upcoming ML adventure. Are you ready for it?

Step one: Notebook

As I said: I hate Microsoft products (except Office 365 — I think it’s f&*’n awesome). So to avoid upcoming frustrations I used Jupyter Notebook on my computer and I encourage you do to so. But if you want to avoid frustrations around Python na Jupyter setup, go grab Google Colab or Azure Notebooks — they both provide online editor and runtime. Enough for basic computation and graph display.

Step two: Bing

If you thought that Bing is dead, you couldn’t be more wrong.
It’s still there on purpose. Our purpose is to grab photos of any type of clothing you want. Fashion mnist consists only 10 types:

1. T-shirt/top
2. Trouser
3. Pullover
4. Dress
5. Coat