You knew this technology would first be used for shopping. Oh yes you did. Riya has launched a visual search engine that allows you to shop visually instead of by keyword. This is a meta-search, with items available from Amazon, Zappos, Lands’ End, and several other merchants. The new search engine is called Like ( http://www.like.com .) It’s in alpha.
The front page has several categories through which you may browse (handbags, shoes, watches) as well as several celebrity pictures. You can click on specially-marked places on the celebrity pictures to find accessories that match what they’re wearing/carrying. You know, if you have an extra couple thousand dollars in your couch cushions.
Start by browsing a category (or clicking a celebrity.) From there you’ll get a grid of items and their pictures. If you see something that looks interesting, you can click on the Likeness Search button beneath the item and get another set of search results based on the appearance of the item you chose.
Sliders at the top of the results screen allow you to specify what’s most important to you: color, shape, or pattern. (An additional slider on the left side of the page allows you to specify by price range.) You can also draw a box around specific parts of an item and rerun a search based just on that area, or click on a color array to specify what color you want.
I did a little experimenting. The sliders seem to work well, as does the general likeness search. The drawing a box around a specific part of the item and getting results based on that was pretty iffy. Drawing a box around specific parts of the item and doing a search based just on shape or color (not both) worked better in the experiments I ran.
Like is currently adding inventory at the rate over 30,000 items daily an plans to add additional shopping categories in the future (including clothing and home decor.) Personally I would like to try this technology with clothing — the handbag, jewelry, etc. search didn’t do much for me. A good idea that worked well for the most part — just needs more categories.
Categories: News