Dear all, thanks for all your help with answering questions and giving feedback over the last couple of months. I'm happy to say that we're finally at a stage where we've hashed 22,452,638 images from Wikimedia Commons and launched Elog.io in public beta: http://elog.io/ Elog.io is an open API as well as browser plugins, that can query and get information about images using a perceptual hash that's easy and quick to calculate in a browser. What the browser extensions allow you to do is match an image you find "in the wild" against Wikimedia Commons. If it can be matched against an image from Commons, it'll show you the title, author, and license, and give you links back to Wikimedia, the license, and a quick and handy "Copy as HTML" to copy the image and attribution as a HTML snippet for pasting into Word, LibreOffice, Wordpress, etc. Our API provides lookup functions to find information using a URL (the Commons' page name URL) or using the perceptual hash. You get information back as JSON in W3C Media Annotations format. of course, the information you get back is no better than the one provided by the Commons API, so if you already have a page name URL, you may as well query it directly, and rely on our API only for searching by perceptual hashes. The algorithm we use for calculating perceptual hashes, which you'll need to query our API, is at http://blockhash.io/ Sincerely, Jonas _______________________________________________ Commons-l mailing list Commons-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/commons-l

This is awesome :) I’m already a keen user! I would have a question. Typically people are really happy when an image they uploaded is reused somewhere on the Internet (or angry, depending on the quality of attribution ;-). We probably cannot scan the Internetz and running every single image out there against elog.io ; but do you by any chance log the successful hits in elog.io ? It would be awesome if we could retrieve the information that elog.io identified File:X at http://example.com/ I must be doing something wrong, but I have not succeeded in using this :-/ This is supposed to be the URL to query per [1] ? < https://catalog.elog.io/lookup/uri> Could you give a working example ? [1] http://docs.cmcatalog.apiary.io/#get-%2Flookup%2Furi%7B%3Furi%2Ccontext%2Cp…

Hi Jean-Frédéric, yes - you're right, I missed that. So sorry!! out No, for that purpose, I think Google Image Reverse Search of TinEye does a fair enough work. Involves a bit of manual digging, but scanning the Internet is indeed something that require quite a bit more resources than we could muster :-) Yes, we do. At the moment, we don't do anything with it though. What we would love to do is to take the information we gather - essentially pairs of images - and run that through PyBossa or similar crowdsourcing platform so that people can contribute to making matches between images by answering the question "Are these two images the same?" If we get more than a few people saying that they are, we can make that connection in the Elog.io database, which could not only trigger some action on "re-use", but which would also make the information about that image pop up automatically when someone browses it without needing to separately click the "Query" button. We do not want to add these matches directly without some validation since that would open up the possibility of abuse. But we think the crowdsourcing could work well, and that in itself opens up the possibility of using the same crowd to gather even more information. Sincerely, Jonas _______________________________________________ Commons-l mailing list Commons-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/commons-l