Search site Like.com goes live today, offering what is believed to be a shopping first — the ability to troll for fashion finds based on the visual content of photos.
Text-based search is comprehensive, but results can be unwieldy. Enter the word “pumps” into Google and 61 million choices pop up, ranging from high heels to hydraulics. With image-based search, shoppers click on a photo they like — say a shot of Tyra Banks in knee-high boots — to find similar boots sold by the 200 merchants in the Like.com affiliate network.
The database has two million stockkeeping units in footwear, handbags, watches and jewelry and 30,000 items are added daily from retailers like Lands’ End and eLuxury.com. A library of 100,000 celebrity images — the inspiration that gets shoppers jump-started — is updated daily.
Related Articles
Apparel will be added to the mix in two weeks, said Munjal Shah, chief executive officer of San Mateo, Calif.-based Riya, which built the search engine. Later, shoppers will be able to upload their own photos and search the entire Web.
You May Also Like
“The big idea here is you’ll be able to search by submitting a photo as your search query,” Shah said. “Everywhere else, you’ve got to type in some words and describe what you are looking for. That works for some things, like an iPod, but doesn’t really work for things that are highly aesthetic or design-
oriented.” Consumers may not have the vocabulary to describe what they want, but they know it when they see it. The site also offers text-based search.
Image-based search demands heavy computational lifting. For instance, calculating the “visual signature” for one photo — including shape, color, pattern and other details — involves 10,000 data points, Shah said. Searches are processed on Riya’s 1,000 computer servers. (Most retailer sites have fewer than 10 servers.)
Riya collects a commission on Like.com referrals that lead to purchases — about 10 percent on average, Shah said. Results are ranked by relevance, however, and not fees paid to Riya.
Founded in 2004, Riya raised $19.5 million in funding (from Bay Partners, BlueRun Ventures, Leapfrog Ventures, First Round Capital) and its staff includes 14 people with doctorates in facial and object recognition and image processing.
Despite the intellectual and monetary might behind
Like.com, its founder and experts agree success will be based on acceptance.
“They are onto an idea here,” said Sucharita Mulpuru, senior analyst, retail, Forrester Research, who previewed the site. “But like all great ideas, it will come down to execution.”
As the volume of online digital content grows, so too will interest in tools to manage it. Google acquired image recognition vendor Neven Vision in August. However, it has not revealed plans for the technology. Microsoft tested image search with Photo2Search, but that has not come to the market.