TinEye – Cool concept – poor execution
On slashdot this morning there was an article about a new website called TinEye. They’ve got a fairly simple but cool concept. Do for images what google does for web pages. You give TinEye an image and their magical search engine goes and finds similar images to what you give it — based solely on that image alone. I signed up and gave it a spin. My results were less than amazing.
I started out by using some of my images on Commons. Specifically, the ones that were used. Every image I tried came back with nothing. So I went over to Flickr and tried an image that I figured would be the most likely to show up elsewhere — again nothing. Then just for fun I went to GOD and tried some images there — again no dice. As a last ditch effort I tried my picture of the Amazon Kindle (from Commons) since I _knew_ that was used on a number of different wikipedia’s and those get lifted all the time. I got a hit!! Apparently some random blogger has stolen my image (Yes, the image is freely licensed, but the licenses require attribution — which isn’t done).
Amused? Yes. Unimpressed? Very. I’m not surprised that nothing from GOD showed up. But Commons and Flickr didn’t show up? If nothing else the originals should have shown up in the search results. Plus, as I stated before, I know a number of these images are used on a variety of Wikipedia’s — I find it hard to believe that not even Wikipedia is indexed. Meh.
I’ll wait till they come out of “Beta”… then at least they should be to a reasonable actually-beta-like stage.