I would like to use images from social media (e.g. imgur, Facebook, Flickr etc) for research (ie not for profit or any industrial use).

  1. Would it be legal to use images (for analysis, etc), if I do not store them locally?

  2. Would it be legal to create a dataset of such images, that I will later publish with my future paper?

  3. If it is not legal, how can there be these existing computer vision datasets of thousands (and even hundreds of thousands) of images taken from Google search or social networks? Surely the authors didn't ask every owner for permission.

  • 2
    There is no such thing as "I do not store them locally". To display an image, your computer downloads it, and only then displays it. Maybe it doesn't stay inside your computer memory long, but it's there (for at least as long as the image is on your screen).
    – user9646
    Commented Dec 9, 2016 at 12:31
  • Of course. What I meant was to ask if there is any distinction between just (a) just using the data and (b) creating a dataset from them.
    – yoki
    Commented Dec 9, 2016 at 16:08

3 Answers 3


In order to copy or redistribute images that you don't own, you have to comply with the terms of the license under which the owners have released the images.

Some images are released under licenses (such as Creative Commons licenses) that allow copying and redistribution under various conditions (e.g. attribution, non-commercial use only, etc. - depending on the specific license.)

For example, the People In Photo Albums dataset was "collected from Flickr photos with creative commons licenses", according to the paper. Similarly, Yahoo! Labs compiled a dataset of One Hundred Million Creative Commons Flickr Images. And Google released an Open Images Dataset dataset of images that are listed as having Creative Commons licenses.

If there is no license accompanying the image or no blanket license applying to all images on the site, you have to assume that it is All Rights Reserved - in this case, you would not be able to redistribute these images.

  • 1. Does Facebook have such a license? 2. In case of all rights reserved, can I still use it for personal analysis and publish results without redistributing? Possibly just link to the owner's image? Thank you.
    – yoki
    Commented Dec 8, 2016 at 22:15
  • While true, I don't think this really addresses the question. The asker isn't asking about republishing images (except for a limited number in their question 2, which they could presumably select to be ones with appropriate licences) - they're asking about performing an analysis on the images. IANAL, but I don't see this as any different to analysing the contents of a book that is in copyright.
    – Flyto
    Commented Dec 11, 2016 at 21:49
  • 1
    @SimonW This answer mainly addresses 2 & 3. Feel free to address 1 in a separate answer (I believe this will depend on jurisdiction, and the terms of use of the websites from which the OP gets the images.)
    – ff524
    Commented Dec 11, 2016 at 21:50

I'm not a lawyer, but I'll go out on a limb a bit here. I would think that your use and analysis of such images is fine, provided that you don't publish any of them in your research and that you don't do anything similar to publishing them. If images are put on the web they are meant to be seen, no matter the license. For you to "see" them, they are transferred to your computer. If you analyze them on the fly and keep only summary information or classification, there is no harm to the rights holder (in almost all cases).

Saving them for later analysis might raise issues later as the "dataset" on which your research was based may need to be preserved. But if your dataset is not the images themselves, but your analysis data points you aren't (likely) in violation.

For research reports and the like, if you need some images, choose only those for which the rights are clear.

Caveat: Copyright differs widely from place to place. Consult a lawyer for edge cases. Me going out on a limb is different from you taking a risk, I realize. European law has become especially strict lately, I note.


For legal advice one should always consult a lawyer directly.

I don't think it is recommendable to (solely) rely on inquiries in a forum or website or even own study of law books as a layman.

Copyright is a very complicated issue and depends heavily on countries' laws. Any advice given here would be opinion based (unless posted by a lawyer) and may have serious financial and professional repercussions.

There are also privacy issues and laws that need to be taken into account.

As a prominent example Cambridge Analytica filed for insolvency proceedings...

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .