[OpenGLAM] One Hundred Million Creative Commons Flickr Images for Research

[Ben O'Steen]

Re: image search. I've found the search not to be comprehensive and when
new metadata is added, the index updates in a unreliable manner. I
regularly harvest the metadata for the images I've uploaded, and have found
distinct discrepancies between what data I know is there, and what turns up
in the search results. For example, a number of the images require
rotating, and Flickr does not offer a way for a user to indicate that to
the account owner. I suggested to people to tag them with "rotate" to flag
them up. A search of images tagged with 'rotate' and belonging to my
account differs considerably from the tag information I know about from my
daily harvests.

For example, this search is meant to show any image with the tag 'rotate':

https://www.flickr.com/search/?q=rotate&m=tags&ss=2&ct=6&mt=all&w=12403504%40N02&adv=1

You may notice that majority of these do not need rotating any longer, and
do not bear the 'rotate' tag. They likely will all have the 'rotated' tag
however. A search for this 'rotated' tag shows a different picture again. I
know that as of 5am this morning, 6462 images have the rotated tag, which
is a different number to what the search gives me.

https://www.flickr.com/search/?q=rotated&m=tags&ss=2&ct=6&mt=all&w=12403504%40N02&adv=1

The search may be better than google's but that isn't saying much. The
search is also built with textual data. I know that Flickr are
experimenting behind the scenes with image analysis to provide other routes
to explore and cluster images together. Imagine if every image on Flickr is
OCRd, or has a field added indicating the three main colours in the image
for example. With images, we should strive to do better than a simple
text-only search.

You are absolutely right to be worried about privacy and licensing
concerns. They seem to be glossed over. They are pushing the responsibility
onto researchers but without making it clear that the images have a range
of licences. Hidden in the T&C for Flickr's sandbox area is a clause saying
that you must not use it to derive any personally identifying information.
Leigh Dodds has captured it and put it up for viewing here:
https://gist.github.com/ldodds/28367dfc533487ea7c5b (line 12 is the key
line)

And yes, Flickr should just let any user upload with a CC0/PDDL type
licence if they wish, a real PITA. We were forced to become a member of
'The Commons' to do so and now we have no opportunity to use a different
licence on subsequent photos uploaded to that account. (It always shows 'No
Known Copyright', regardless of the license I try to apply.)

Ben


On 25 June 2014 16:35, Laurel L. Russwurm <laurel.l at russwurm.org> wrote:

>  Perhaps I just don't understand the point of this, but it seems to me
> that Flickr already has an image search capability at least as good (if not
> better than) Google's... and I suspect Google has done something to limit
> the incidence of Flickr returns in its own search method.
>
> My privacy hackles are raised by:
>
> "the task is to build a system capable of accurately predicting where in
> the world the photos and videos were taken without using the longitude and
> latitude coordinates."
>
> Guess its time to only limit cc images purged of metadata or maybe to stop
> posting cc images to flickr at all.
>
> Personally I think it would be much more useful and cost Flickr much less
> to add the capability for users to post CC0 and public domain marked
> images.
>
> Regards,
> Laurel L. Russwurm
>
>
>
>
>
>
>
>
>
> On 06/25/2014 05:04 AM, Ben O'Steen wrote:
>
> An excellent initiative from its description :) It will be interesting to
> see how this actually pans out.
>
>  Ben
>
>
> On 25 June 2014 09:35, Johan Oomen <joomen at beeldengeluid.nl> wrote:
>
>> Good morning (for those in Europe;-),
>>
>> I just came across this announcement from Yahoo Labs: "One Hundred
>> Million Creative Commons Flickr Images for Research”.
>>
>> More information here:
>> http://yahoolabs.tumblr.com/post/89783581601/one-hundred-million-creative-commons-flickr-images-for
>>
>> Best wishes,
>> Johan
>> @johanoomen
>> _______________________________________________
>> open-glam mailing list
>> open-glam at lists.okfn.org
>> https://lists.okfn.org/mailman/listinfo/open-glam
>> Unsubscribe: https://lists.okfn.org/mailman/options/open-glam
>>
>
>
>
> _______________________________________________
> open-glam mailing listopen-glam at lists.okfn.orghttps://lists.okfn.org/mailman/listinfo/open-glam
> Unsubscribe: https://lists.okfn.org/mailman/options/open-glam
>
>
>
> --
> Laurel L. Russwurm, Author <http://laurel.russwurm.org/blogs/> § about.me
> <http://about.me/laurelrusswurm> § Libreleft Books <http://libreleft.com>
>
> _______________________________________________
> open-glam mailing list
> open-glam at lists.okfn.org
> https://lists.okfn.org/mailman/listinfo/open-glam
> Unsubscribe: https://lists.okfn.org/mailman/options/open-glam
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/open-glam/attachments/20140625/ab093608/attachment-0003.html>