Riya, the face detection or image search company is trying a new approach, which Michael Arrington likes. The face detection idea didn't fly so now they are trying "likeness" search targeted at the fashion industry and have renamed the service Like.com. 
Robert Scoble thinks it is really cool too. I am skeptical based on my past experience.
The idea is to train the system with a catalog of pictures of movie stars to recognize fashion accessories, and to find "like" versions of them for sale on the Internet. Riya believes the fashion industry is more tolerant of mismatches and will actually enjoy the serendipity of "like" items.
Most of you know I was director of engineering at AltaVista almost 10 years ago where my team focused on multi-media (photo, video, music) search. We tried face detection, skin detection, and likeness search and did a pretty good job. There were two problems; 1)the technology wasn't accurate enough, and 2) there wasn't a monetization or business model for it. Riya found out that things haven't changed much. I could have told them that, in fact I predicted it wouldn't work in my initial review almost a year ago entitled "What problem does Riya solve?"
Liz Gannes at GigaOm asks "So how well does it work? Riya CEO Munjal Shah showed me a bunch of searches that worked really well, like strappy red high heels and a busily patterned bag, but when we got off-script with a shiny necklace, the results were much more unimpressive. Since the site went live tonight, I’ve had more trouble than I would expect finding a search that returns high enough quality of results to get a screenshot. I do think this is a cool idea but it’s not dazzlingly good yet."
Dan Farber interviewed Riya CEO Munjal Shah and got the following information, "Like.com is starting out by indexing images from jewelry, handbag, shoe and watch merchants, and will add clothing soon, Shah said. Jewelry, handbags and clothing are about a $15 billion business online. The target audience is women from 20 to 30." "Like.com gets paid via cost per click and cost per action. The company is looking to advertising as a revenue source, but is focused on building the audience first."
My take? Riya has a much better chance of success focusing on the fashion market because "likeness" is OK, exact matches are not necessary. Secondly, the Cost Per Click (CPC) business model will be much more effective than consumer based advertising. That addresses the two concerns I had initially.
The lesson for entrepreneurs is don't have preconceived notions about how your product/service will be used. Test with lots of different customers to discover where they see value. Remember, it is not about the technology...it is about the problem it solves.
Subscribe - To get a FREE automatic feed of all future posts subscribe here, or to receive them via email go here and enter your email address in the box in the right column.
Firstly, i tried to use like.com and i found it very confusing to use and moreover it did not find relevant items .. so for scripts it might be working fine but not when it goes off the script.
Secondly, like.com will have a hard time changing shopper's habit by making them go for shopping through their website following their process. I presume they will license this technology to retailers for them to put it on their own website and have a small "similar items" button with each item.
but again maybe i am wrong ..
Posted by: Raza | November 08, 2006 at 11:04 AM
Ich habe dieser BLOG gelesen und ich finde ihn interessant.
Bitte besuchen Sie uns auch-
http://www.fashionbags.de
Posted by: jameshkoler | November 09, 2006 at 05:57 AM
Check http://www.imgseek.net/ for an opensource desktop application with content based image search and its server side version: http://server.imgseek.net/ (may be interesting for those that want to integrate visual search into their website or application)
Posted by: Paul Pruitt | November 12, 2006 at 09:50 AM