TinEye: Image Searching

I don't often write about technology on this blog, but TinEye seemed like something both important and fun to talk about since it's related to images, and by extension libraries and art. As we all know, searching for images on the Internet can either be easy or challenging. It depends on how you describe something and, more importantly, how it's been described by the host. Almost all of us use Google Images for most of our image searching. You type in a few words and a page of thumbnails appear. The likelihood is that some of the images are relevant; however, not every work always is an accurate hit. The reason why this happens is because search engines are looking for keywords near the actual image on a website. They're not actually searching the image itself. Admittedly, Google is doing an excellent job of enhancing their image search capabilities by incorporating the Google Image Labeler project, which invites people to participate in tagging digital images for better searching. But, TinEye just may be the future of image searching. It doesn't use text to search, but actual images. You upload an image to TinEye or point it to a particular website with images. The program "reads" the image using an algorithm. It then searches using this algorithm for other images on websites they have indexed. Any matches--including derivative versions of the image--are generated in a list that allows you to click to the website to see its context.

It's pretty neat, but what are the implications for its use? In other words, if you're looking for an image of something, TinEye probably isn't going to help. But if you already have an image and want to find better quality versions, other places where the same image has been used on the Internet, or derivative versions of an image, then TinEye is the way to go. I can see it eventually becoming integrated with Google Images, where you do a basic text search, and if you find an image, you could click on the TinEye link that will allow you to "Search for more images like this one" to generate the type of results you really want. Someone at work suggested to me that it could also be integrated with digitized books. TinEye would then allow one to track everywhere an image has appeared in print. For artworks protected by copyright, it would assist artists and estates in ensuring proper use of an artist's works.

If you want to learn more, check out its functionality by going directly to http://tineye.com/ and registering for free to try it. They also have some fun widgets like the one above, Michelangelo's Creation of Adam from the Sistine Chapel ceiling. Just click on it and you'll see a random sampling of other websites where this iconic image has appeared in many fun-filled ways.