Photo/sketch based challenge?

[Matthias84]

Hi, as Wikipedia captcha points out, image recognition seems to be still a challenge for robot based attacks.

So my question is, if we could offer a challenger that makes use photos of different categories e.g. [cats, dogs, birds, faces, horses, ...] and asks the user to identify a single image of a picked category within a mixed set of images? I saw similar approaches already for Mediawiki.

As it might be superb to avoid third party services offering this images, I suggest to make use of this categorized archives:

• OpenClipart
• Wiki Commons
  and then maybe altering it. But you need to pay attention to the copyright and approapriate attribution of course (without offering a new attack vector ;))