Photo recognition startup Riya has dramatically changed course, and is setting up shop today at Like.com as a visual search engine for jewelry, handbags, watches, and shoes. The land of women’s fashion is a bit of a leap for most any Silicon Valley company, but Riya does have the big advantage of jumping into a demographic that is not polluted by redundant tech startups pitching their wares.
Sure, there are lots of shopping search engines (some of whom we’ve written about lately), and an increasing number of them, such as ShopWiki and Become.com, do color search. What Riya is doing is using its photo recognition technology to help shoppers find products that are similar in ways that are harder to input or describe, such as the strappiness or heel of a sandal, or the pattern of a fabric.
For now, the way this works is users can page through pictures of celebrities (kept up to date through a deal with Getty Images), then zero in on a particular bracelet or handbag they like and find similar products for sale (from about 200 online merchants who have signed up for affiliate programs). Users can also input text descriptors like “flower pendant” or choose from categories to get a search going.
The cool parts of the implementation reveal themselves once you’ve started a search. You can refocus any query by selecting an item and dragging the cursor over the part of an image you really like. After you tell the search engine to focus in on either the color or shape of the swatch, it redoes the search to your liking. There are all the best-of-breed features from other vertical search engines — sliders, dynamic checkboxes for narrowing to certain brands and stores that come up in a search, and “breadcrumbs” done visually display using thumbnails to show your search path.
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. We also asked about matching accessories across outfits and inputting our own pictures as search queries — Shah assures us that these features, as well as social tools and expanded categories, are coming soon.
Shah admits that part of the allure of the new product is “the threshold of success is lower.” The claim is, shoppers don’t mind a slightly different handbag than the one Scarlett Johansson was carrying at that movie premiere last night; they find it interesting. With facial recognition, there’s no room for error. (Riya has left its facial recognition search engine at Riya.com, though it has been neglected in favor of the new product for the last eight months.) For those who want to dig a bit deeper, Shah, who is not a bad storyteller, has been chronicling his strategy shift in a series of posts on his own blog.
It’s lucrative market niche. However, the relevancy of what you find to what you search for is the key here.
Liz – Like isn’t perfect but we have found that it improves in accuracy as we add more inventory as the likelihood that something similar to you item exists, increases. We will also tune the algorithms between alpha and beta. Keep the feedback coming.
I have a couple of videos with Riya (an interview with Riya’s CEO and a demo video) over at http://www.scobleshow.com — linked to from my blog at http://scobleizer.com/2006/11/08/riya-reborn-is-really-cool-way-to-search/
great one. riya should also revise its buying process.
munjal. after google not acquiring riya, Riya have kept growing and is growing like anything.
keep it up!
Nice article. Riya have some competition on this one – London based PIXSTA launched http://www.chezimelda.com recently – a huge online shoe store browsable by image. Comparing the two, Like.com’s site is more polished and great fun (I like the celebrity section), but for a PURE shopping experience http://www.chezimelda.com is much more intitive. It will be interesting to which one consumers go for.
liz,
its useless to be frank. Unless they are able to federate ads to websites, blogs etc.
Riya’s whole premise is that its a search engine ? like.com doesn’t act like a search engine..so what is it ? are they competing with shop.com or images.google.com…or froogle.com…i don’t know what all this hype all about..
It’s tip top. I’m just not sure about the colour.
Having tested both Like.com and ChezImelda.com (also this one is only for shoes) following Steve’s comment – ChezImelda is claiming to have 90,000 shoes indexed by image similarity. My findings are. I find the interface from Like.com just to much. Easy browsing like on ChezImelda is much more my style. Quick click on an image the network resorts itself and I have a view into a large base of shoe inventory. Reminds me a lot of how simple google is and maybe also a reason to be more successful. Just compare and report back! would love to see other people views.
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)
This article born a lot of ideas in my head – thank you very much for this article Liz.
Hi Liz,
You’ve highlight a great thing about Riya. Keep psoting such a good article.
Cheers,
http://magnetic-jewelry-clasp.blogspot.com
riya should also revise its buying process