EditorialsFeatures

flickr retrievr and visual search tools

By Brent Phelps January 05, 2006

Comments (0)

It seems that an area ripe for improvement in the stock photo industry is the development of new and accurate search methods to help customers connect with the right image. Traditionally the way to search for an image is based on metadata associated with an image such as the caption and keywords, but this method does not take into account many of the visual characteristics of the image such as composition and color. This also presents problems when you consider the idea of searching multiple collections at one time in which different keywording and search standards make it very difficult to display the search results in a meaningful way. For these reasons we have seen many companies make attempts at incorporating other more intuitive visual search features such as a search by color feature or iStockphoto’s CopySpace search feature. These features are typically incorporated with a metadata search and, for the most part, seem to work moderately well but it is clear that there is much to be learned about developing a true visual search. That said, it is great to see new innovation and I recently came across an interesting experiment created by System One called the "flickr retrievr".

The retreivr tool is a prototype search method that allows one to search for images on flickr in a visual way rather than searching by keyword. To find images the user simply uses the tool to draw a rough sketch of what he or she is looking for. The creators are quick to point out that this is a prototype and the result can be useful but not always entirely accurate. Another thing to keep in mind is that retrievr doesn't do object/face/text recognition of any kind, so if you're drawing an outline sketch of a chair, it almost certainly won't show images of only chairs. However, as a prototype, the tool is fun and interesting to play with and if it were refined and perhaps coupled with a good metadata search this feature could prove to be a step closer to a true visual search. Only time will tell whether visual search will revolutionize the industry but it seems clear that once fully developed a visual search could dramatically alter the way we search for visual content on the internet.

Post a Comment


Name required

Email required but won't display

URL posted with nofollow attribute

Your Message

Remember my personal information

Notify me of follow-up comments?

Please enter the word you see in the image below