Mobile image recognition technology identifies objects like shoes, handbags, and golf clubs in images, videos, or directly through a tablet or smartphone’s image sensor. The technology may make a new form of mobile commerce possible in just the next few years, allowing shoppers to see products in use or on another person and instantly buy those items from a mobile application.
Imagine walking on a busy street in San Francisco or Boston. You see a fashionable person wearing a attractive suit or carrying a beautiful handbag that you’d like to have for yourself. You pull out your iPhone 7 — this technology might take a few years — and a mobile image recognition (MIR) app instantly identifies the suit or handbag, offers prices from three or four online retailers, and provides a one-touch “buy now” button for you to complete the purchase.
This sort of “snap shopping” is feasible, given the current state of MIR technology and the growing ubiquity of mobile devices, including tablet computers, smartphones, and even Google Glass.
Current State of Mobile Image Recognition Technology
Today, MIR technology may already be used in mobile applications to identify particular advertisements or even specific products in a known context.
For example, iTraff Technology offers an application programming interface that allows a mobile app developer to upload known images of products or a billboard ad and have those images recognized when a user snaps a photo of them with a mobile device.
The solution is limited, but it is also a clear forerunner to providing potential customers with a significantly more interactive view of the world around them, including the possibility of making everything you see, buyable.
One could think of it as “showrooming” everywhere.
Similarly, Kooaba has technology that can make advertising — even in print media — interactive. In concept the next step toward recognizing the objects seen every day may only be a matter of small improvements in technology and a large image database.
LTU Technologies, which was founded by scientists and researchers from the Massachusetts Institute of Technology, Oxford University, and the French research body INRIA, is also making MIR solutions that could be applied to mobile commerce.
“Already, mobile visual search is being used by retailers in Asia and Europe for targeted m-commerce applications, and within three years, we will see rapid adoption of this technology by U.S. retailers,” said Stephen Shepherd, LTU Technologies general manager in an interview with Forbes magazine earlier this year.
Add to this list of MIR technology examples, Google Goggles, which has been available for nearly three years. It recognizes real world items including products that can be purchased online.
MIR May Offer Limited Results
Significant concerns for small or mid-sized online merchants may include who will supply the most popular MIR applications and how many merchant results an MIR mobile commerce application might show in response to a visual query.
When a shopper goes to Bing or Google on a desktop computer and searches for running shoes, both search engines return an almost unending list of results, so that even a small specialty shop will eventually show up. These pages also have long lists of pay-per-click advertisers, so small businesses can buy their way onto a page if the competition for organic results is too strong.
On a mobile device, there are relatively fewer listings visible when a search-engine-results page loads, but other businesses can still be found with a swipe or two.
However, with MIR it is possible that only one-to-four retailers might be shown in response to an image query, potentially limiting how well small businesses could perform.
Likewise, if the most comprehensive and popular MIR mobile commerce applications are ultimately branded by individual stores, it is possible to be shut out completely. A MIR app from Amazon or Walmart would return results from Amazon or Walmart not from all available merchants.
Some have also suggested that the folks using MIR mobile commerce applications would choose preferred merchants. So that visual query results would favor particular sites or sellers. If this were the case, online stores would need to promote services and prices, while encouraging inclusion in a manner similar to the way that some businesses seek Facebook Likes.
How Small Online Retailers Can Compete for MIR Commerce
If in fact, image-recognition does drive a new breed of mobile commerce, small businesses may have at least three options for being competitive.
First, small shops could work together to offer an MIR-based mobile commerce application that fairly returned results from all participating businesses. These shops might be organized around industry associations, buying groups, or even a software platform. As an example, Volusion, Shopify, LemonStand, or Magento might offer a shared MIR mobile commerce application. Potential customers using the app would see results from the pool of small and mid-size merchants using the associated platform.
Small business, particularly those in specialty industry segments, could produce their own MIR-powered mobile commerce applications. A retailer of golf clubs could offer an app that would recognize a particular nine iron or driver as it is being used by some other player in a shoppers’ foursome. Or an online retailer of automotive products might offer an MIR-based application that recognized vehicle makes and models so that shoppers could order auto parts or even see appropriate repair tutorials.
Relatively small online merchants might also be able to compete for visual queries with a multichannel ecommerce approach, offering products on marketplaces like Amazon, Ebay, Sears, or Newegg or advertising on price comparison engines. Assuming that one or more of these marketplaces or price comparison sites begins to offer an image recognition application, the merchant could enjoy some sales via the marketplace’s development efforts.
Problems with MIR Technology
For MIR technology to truly take off and enable a new sort of mobile commerce, image recognition firms face three important problems.
MIR solution providers will need to address the technical challenges of storing hundreds of thousands (if not millions) of reference images in some cases and quickly returning accurate results from a reference image database.
These companies will need to manage privacy concerns, since many people — however stylish — won’t be pleased with random folks snapping photographs of them whatever the reason.
Related to the privacy concern, MIR solution providers need to reset or adjust social norms so that taking out a mobile device and snapping images or even recording video of individuals somehow seems less creepy.