Yeah, I definitely submitted this a bit early. I've added a fair amount of icon detection (no more search icons or books with question marks) and I've overhauled the interface so it's ajaxy and a good bit more fun to play. Thanks for the feedback!
I have been considering using images from wikipedia in a project that is much different than this but I was wondering: is it okay to embed the images so that they load from wikipedia, or would you need to download them and save them to your own server? This might be called inline linking?
Nice. May want to rule out images that are too common (the search magnifier) or routine/small (nationality flags, though I got most of those right).
Also, especially after answering, I'd like to click through to the real article to see the image in context -- so it's be nice to provide that link.
It might be interesting to use the same corpus in a different challenge -- for example, five articles and five pictures, match each to the right article.
That's a great idea. Right now it's using Wikipedia's random article generator (http://en.wikipedia.org/wiki/Special:Random) to drive content, but it would be cool to group by a given category.
11 comments
[ 3.6 ms ] story [ 37.7 ms ] threadPerhaps set a lower limit on image sizes that get used?
I'd still suggest a lower limit on image sizes. I'm seeing icon-sized pictures of the Golden Gate Bridge show up, etc.
Also, especially after answering, I'd like to click through to the real article to see the image in context -- so it's be nice to provide that link.
It might be interesting to use the same corpus in a different challenge -- for example, five articles and five pictures, match each to the right article.
http://en.wikipedia.org/wiki/Category:Lists
And please filter for icons.