It started as a need for a dataset with images and turned into writing a tool which can download parts of Imagenet using its API. I also created a little analysis of the current state of the ImageNet URLs from its API which helped to significantly improve the downloader.

Here I wrote about the process: https://mf1024.github.io/2019/06/09/how-to-scrape-the-imagenet/

I hope it’s helpful to others and comments are appreciated.