Big Data Helps Brands Take On Logo Photoshoppers
It's no surprise that the Internet is a free-for-all of opinions, rants and raves. For a global organization that has invested millions of dollars and many years of effort building a brand, negative comments on social media sites, blogs and other online forums can be damaging. And while standard search tools are often effective for monitoring text-based posts, potentially harmful images -- a defaced corporate logo, for instance -- are trickier to find.
- The Critical Importance of High Performance Data Integration for Big Data Analytics
- Speed to Solution: Are There Shortcuts to ROI?
White PapersMore >>
New York-based LTU Technologies, a small image-recognition subsidiary of Japanese IT service company Jastec, thinks it has a solution. Its image-search technology takes a big data approach by scanning petabytes of image data to find both positive and negative uses of brand images.
"Most companies have millions in dollars of brand equity in their visuals," said Stephen Shepherd, general manager of LTU Technologies, in a phone interview with InformationWeek.
Vast numbers of brand images are posted regularly on Facebook, Twitter and other social media services, Shepherd noted. And while many of these images may show a product or brand in a positive light, some are quite critical and potentially libelous.
"If (companies) find their visual identity has been expressed in a way other than they intended, they like to know about that," said Shepherd. "Not so much to sue, but to do a cease-and-desist and get the people to stop."
[ Want to break into the big data game? See Big Data Analytics Masters Degrees: 20 Top Programs. ]
LTU Technology's image recognition software dates back to 2000, and is based on core research from the Massachusetts Institute of Technology, Oxford University and French technology company Inria. Initially used by government agencies to locate missing children and stolen art -- and to perform other classified tasks the company isn't at liberty to discuss -- the technology has been "refined many, many times" over the past decade, Shepherd said.
According to an LTU whitepaper, the firm's recognition software is designed to function much like a human eye, "memorizing" subtle differences in image colors, shapes and shades. It also studies the subtle relationships between lines, shapes, curves, shadows and text.
"It breaks down the pixel content into a binary fingerprint," said Shepherd. "We call it an LTU DNA. Each image has its own identity, its fingerprint."
The image recognition engine automatically searches and compares millions of images online, spotting minute differences and generating a report if it finds brand logos and related images that appear to be altered.
"Some of these large corporations have very strict brand standards," Shepherd said. "Their brand identities are huge and spread out all over the world."
The whitepaper includes examples of defaced corporate logos found by LTU's image recognition tools, such as petroleum giant BP's logo melting into what appears to be a toxic oil spill (complete with fish skeletons and dead birds), the Starbucks mermaid assuming a suggestive pose and an upside down McDonald's "M" logo with the words "Weight, I'm gainin' it."
"A Starbucks logo that's been intentionally defaced is information that would be relevant to their corporate communications department, and to store management," Shepherd said.
Critics might contend that there's a creepy, Big Brotherish vibe to multinational corporations using image recognition to bully its critics into silence, but Shepherd pointed out that companies could seek a less threatening solution as well.
"They could reward and re-engage consumers that took it upon themselves to communicate their experience -- positive or negative -- through the image," he suggested.
Shepherd noted that, in addition to major corporations, LTU's potential customers include reputation management companies that scour the Internet for harmful information about their clients.
LTU's technology has other uses as well. One American clothing brand, for instance, used the image recognition software to uncover large amounts of counterfeit merchandise that used its brand name and product style, according to LTU.
Attend Interop Las Vegas May 6-10 and learn the emerging trends in information risk management and security. Use Priority Code MPIWK by March 22 to save an additional $200 off the early bird discount on All Access and Conference Passes. Join us in Las Vegas for access to 125+ workshops and conference classes, 300+ exhibiting companies, and the latest technology. Register today!