Imagine conducting an image search whereby the search engine uses features from an image to retrieve similar images. If you thought to yourself “That’s already something Google offers”, you’d be right. However, they aren’t the only ones tinkering with similar image search technology. Last year during TechCrunch 50, Gazopa launched in private beta, meaning 99% percent of us couldn’t get in. Now, the rest of us peons finally have the chance to check it with Gazopa announcing that they’ve moved into open beta.
The way that Gazopa works is it allows users to upload a picture, enter a URL of an image, create a drawing, or right click on an image anywhere on the web (using a plug-in) and it will then find similar images. You can even search for videos this way, and all you need is a thumbnail of a video to search for similar videos.
The search results are then filtered, mostly by analzying the color and shape of the object or person in the image. So, for example, you can give Gazopa the URL of a picture of a yellow truck and it will find pictures of similar yellow trucks online.
Some of the other cool things that you can do with Gazopa include:
- Search Flickr photos, and tell Gazopa to sort out those that lack a Creative Commons license.
- Find images related to the latest news, which can be filtered by time, shape and size.
- With the Gazopa iPhone app, you can take and upload photos with your iPhone to get similar images off the web quickly.
More than 40,000 users have tested out the service since September 2008. It’s still a little buggy at times, but sufficient for still being in beta stage. If it were a full launch and coming out of beta, well that’d be a different story.
Gazopa, which is owned by Hitachi America, competes with Google in similar search, but according to project leader Hideki Kobayashi, Google doesn’t allow users to find similar images of all images displayed and that uploading a picture by yourself isn’t possible. Another competitor of Gazopa is TinEye. TinEye doesn’t specifically look for “similar” images, and instead tries to find exact matches instead.
It sounds like a promising endeavor, but I’m still not sold on it. To me, these kind specialty searches will always be a bit of a novelty, rather than something I’d first run to when I want to find an image. Let us know what you think about it in the comments section.
