Reuters Insider - Snap, pick, pay: Visual search could change the way you shop

Intelligent Protection Management

31/10/2014 20:17

Click the following link to watch video:                              
 https://insider.thomsonreuters.com/link.html?cn=share&cid=1322494&shareToken=Mzo0MDQ0MTIwZi02NjE1LTQ5NjUtODRmZC1jNDdmZWNkYzkxNGQ%3D&playerName=ReutersNews 
                                                                       
 Source:             Thomson Reuters                                   
                                                                       
 Description:        Retailers like Amazon and Target are rushing out  
                     mobile apps that allow you to snap an image of    
                     what you want to buy. Fred Katayama reports.      
 
 
(To access all exclusive Reuters Insider programming visit: http://insider.thomsonreuters.com) 
 
 Short Link:  http://reut.rs/1paQt9q  
 
 
Transcript (May be auto-generated)

 Number Four, by Jackson Pollock from 1951- Even a thousand words sometimes just 
can't paint a picture. So you don't know Jackson Pollock but you need to find 
out who drew this. Describing it is a pain. To the rescue, comes visual search. 
Google Images can give you the answer on your desktop. But now with the advent 
of mobile apps, you can use visual search for shopping. Adi Pinhas founded the 
visual search startup Superfish. The world is a very visual place and we have 
visual questions all day long. We see stuff around us and we want to learn who 
did that, is there similar things, what can I learn more about it, and so on. 
Big retailers are rushing up mobile apps where you can snap an image of what you
want to buy. The Firefly feature built into Amazon's new smartphone recognizes 
70 million products like Altoids mints and gives you the option to buy it on 
Amazon. Target's test app, In a Snap, allows you to take pictures of products 
featured in select magazine ads and Target catalogues. But there are 
limitations. You're mostly restricted to shooting images of labeled items like 
books or flat well-defined objects. 

So forget that black puppy. And you could only buy things off their websites. 
Enter Superfish. Its technology recognizes 3D objects. Take its app PetMatch. 
Here at Madison Square Park in Manhattan, I've come across this cute little 
Boxer named Lily. I'm going to pull out my app, take Lily's picture - say, 
"cheese". PetMatch's algorithms rummage through its entire database and, within 
seconds, up pops up a series of photos of dogs that resemble Lily - all 
available for adoption in the New York area. And Superfish's Like that Décor 
app does ditto with couches and other furniture - hunting through its database 
of more than 30 million products at 5,000 retailers to suggest similar items. 
Google says the next hurdle is finding a way to search through video, taking 
advantage of the multiple frames captured for each object. Forrester Research 
Analyst Sucharita Mulpuru says 10 years could pass before visual search delivers
results consistent enough for it to become mainstream. And it's a very complex 
computer science problem because what we're essentially asking is that a 
computer- an algorithm recognize an image from any direction under any lighting 
circumstances, and presuming that the photo may not be a perfect photo either, 
it could be taken from a distance or the pixilation may not be that precise. 
Over the next six months, Superfish plans to launch shopping apps for shoes, 
jewelry, and gardening. Eventually, the visual search is going to be a part of 
the platform. The visual search is going to be in every browser. It's going to 
be a part of the camera in every mobile. And then it will do that. It will do 
everything. One day, shopping could become as easy as snap, pick, and pay