TinEye logo

TinEye logo

Image search is hard.  Web pages are full of text, links and metadata that tell you about them – they lend themselves to being searched.  Images on the other hand mostly just come with a filename and if you’re lucky a caption – that isn’t a lot information to go by.  As for the image itself, getting a computer to understand a page full of text is a lot harder than getting it to understand a picture – image recognition is still in its infancy.

So along comes TinEye with a different way of looking at image search – instead of entering text that describes the picture you’re looking for, you give it a picture to look for.  Now obviously this isn’t much use unless you already know what you’re looking for, but it’s fantastic if you’re trying to find the original source of an unattributed picture.

Reblog this post [with Zemanta]
blog comments powered by Disqus