For machine learning purposes, I need 10,000 random photographs from Wikimedia Commons.

I have found Imker to download all files within a category, but nothing for random pictures.


  • Gratis
  • Really random, not all items whose name starts with "A" for instance
  • Only photographs, not PDF/etc, though I can also filter out afterwards
  • I think you are going to have to scrape the meta/directory pages, build a tree of categories & subcategories, then randomly generate a URL :-(
    – Mawg
    Commented Sep 20, 2018 at 13:08
  • 1
    @Mawg: commons.wikimedia.org/wiki/Special:Random/File might help to some extent.
    – Nicolas Raoul
    Commented Sep 21, 2018 at 0:51
  • I think that is 90% of it. Just check if it contains an image, and request another if not. Are you concerned about the randomness of it?
    – Mawg
    Commented Sep 21, 2018 at 6:52


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.