Marketers look to "facial coding" program

The new technology that can read your face, and your mind (Yahoo - Australia)

Using the camera in your personal device, the new facial recognition software interprets your emotions by mapping the minute movements of your eyes when you smile, your mouth when you gasp or the furrow of your brow when you are confused.

“They use the technology to make sure that people have understood their content,” Matt Newcomb from Proctor and Gamble said.

Some (me included) might quibble that this isn't technically a facial recognition system because it isn't used to uniquely identify an individual within a larger set of individuals even though true face-rec and this emotional response detector use some of the same underlying technologies.

Some organizations that opt for this type of technology will do a better job of explaining its use than others.

There’s a fine-line between harvesting personal data to assist or to exploit

How To Collect Consumers' Data Without Freaking Them Out (Fast Company) — Five tips for brand leaders to consider when harvesting personal data so consumers feel okay about giving it up.

A discussion of facial recognition and advertising

Consumers Have No Constitutional Protection From Facial Recognition (Internet Evolution)

Over the past decade, marketers have increasingly relied on facial recognition technology (FRT) to create personalized advertisements. FRT depends on complex algorithms to identify a person by measuring the size, angle, and distance between a person's facial features. FRT then uses this information to search a database of similar features and matches the image to a stored reference photo. Within seconds of capturing an image, FRT can detect and identify a single person in a crowded public area.

Unlike earlier discussions of the use of facial recognition technology used to ascertain demographic characteristics rather than a unique identity, this article discusses true facial recognition in advertising.

Other posts containing longer comments on demographics vs. identity with respect to facial recognition:
Burgeoning Facial Recognition: How come no pitchforks?
FTC Freestylin' on Face Recognition

UNICEF awareness campaign for universal ID in Paraguay

In what has come to be known as the "No Name Match," UNICEF brilliantly harnessed the power of the Paraguay-Uruguay World Cup qualification match to drive the issue of universal ID for Paraguayans to the front of people's minds before recent elections.

The two-minute video below is really good.

 

Without universal legitimate ID, it's harder to make a lot of other things work that most of us take for granted. Universal vaccination against preventable disease, compulsory primary education, effective social safety nets — all of these things get a lot easier if everyone can prove their unique identity.

Face rec for quality assurance

Edinburgh Airport installs biometric system to track passenger movements (Computerworld UK)

An anonymous facial image is taken of each passenger as they check in and the time it takes each to reach certain waypoints plotted over time. If this time breaches a pre-set parameter for enough passengers, alerts can be generated.

The principle is that moving passengers from check in to the terminal increases their satisfaction with that airport and boosts the amount of time they have to spend money in the retail outlets that generate profit for airports.

The system can also be used to track the movement of passengers through the airport as a whole.

This is another really interesting application for facial recognition technology and, unlike other uses of face technology better described as demographic detection, this one actually is a true face recognition application.

Although it is a true face recognition application, it isn't really an ID application so long as the facial image taken at the time of passenger is not linked to other personal information and it is deleted after the person reaches the "finish line."

The item of interest to airports in this case is the length of time it takes real individuals to travel through various points between check in and the jetway. It's a more sophisticated measure than a simple count and real-world measurement wasn't easily automated before face recognition technology.

Airports in the UK have been early adopters of face recognition for this application because they are held to certain performance metrics (and subject to fines) for airport throughput. Having accurate real-time information on passenger flows can inform on-the-fly staffing decisions. For example, additional security screeners can be dispatched in the event a slow-down is detected, saving passenger time and the airport money.

Though airports have been early adopters, this basic application has obvious utility in shopping malls, department stores, planning for emergency evacuations, and large facility scheduling.

SecurLinx has experience in the design and deployment of this type of system. Our FaceTrac system is readily adapted to the challenge of on-the-fly enrollment, finish-line matching, reporting, and automatically purging image data.

Face recognition, marketing and privacy

It's Your Face. Or Is It? (Press Release at Marketwire)

"From a marketer's point of view it's heaven. They can tailor ads, products, even prices based on your age, tax bracket, social media persona and purchasing habits. Marketers will pay handsomely for that information." For example, NEC has developed a marketing service utilizing facial recognition technology. It estimates the age and sex of customers, along with the dates and number of times that customers go to each store. This information is then analyzed to help predict trends in customer behavior and shopping frequency.

"From a consumer's point of view this could be a nightmare -- the ultimate invasion of privacy."

Johnson continues, "I'm not just a brand strategist. I'm also a consumer. And I'd like to speak with the voice of reason. New technology can offer enormous benefits. It also comes with enormous responsibility." Johnson firmly believes we are collectively charged with that responsibility. We have to ensure this facial recognition technology does not become an all out assault on our privacy. "Do we want our children to be added to these facial databases? Probably not. Do we ourselves want to be added without our knowledge or permission? Probably not."

We tackled the very interesting topics of marketing and the privacy of faces in this post from 2011.

It's also worth noting that there are two different ways facial recognition technology can be applied to marketing in the bricks-and-mortar world. True face recognition matching a face to a unique individual so as to send a marketing message tailored for that one person is still pretty hard. Inferring demographic traits of a person by using facial analysis technologies does not rely on a unique identification and may provide a bigger bang for the buck (ROI) than true facial recognition.

Facial recognition for better television ratings

Nielsen Explores Facial Recognition Tech For Ratings

Privacy issues remain, acknowledged Fuhrer. But given the ubiquity of facial recognition-equipped devices, he adds: “There’s a tremendous amount of acceptance” compared to just a few years ago. And there’s also a “layer of anonymity” that can be applied to any facial analysis system deployed by Nielsen, Fuhrer said.

The Neilsen Company has been the gold standard for television ratings in the US. Now, they're contemplating the application of facial recognition technology in televisions that support it in order to get more detailed information about the viewing habits of those included in their survey.

Nielsen pays its sample and is really transparent about the information is collects from individuals. My guess is that the biggest new privacy issue at work here is the potential abuse of non-Neilsen information gained by having access to a working video camera in each of the new smartTv's.


UPDATE: See also Unilever, Coca-Cola Utilize Facial Analysis To Enhance Ad Tests

Face Rec System Can Sell Lipstick and Bust Terrorists

I guess, in a few cases, it could do both.

From Testing Lipstick to Spotting Terrorists (IEEE SPECTRUM)

Talking with Robin during his visit to Palo Alto, Calif., last week, he definitely seemed like a man tugged in two directions. While he was happy to talk about the successes of the technology in security tests, he kept bringing the conversation back to its applications in department stores, guiding women to selections of hair color and makeup.

Get me rewrite.

Very Odd "Facial Recognition" Article at smartplanet.com

Two things jumped out at me while reading San Francisco bars: Buy a drink, become profiled by cameras by Charlie Osborne at smartplanet.com: the scare quotes around forms of the word 'anonymous' and a novel formulation of privacy.

The scare quotes are here...

Venturebeat reports that Chicago-based startup Scenetap has combined “anonymous” facial recognition technology in venues with mobile technology so socialites can choose where next to go on a Friday based on their preferences — all provided through cameras in different venues.

...and here...

Scenetap promises the technology collects data “anonymously” and nothing is recorded or stored, and it is based on sophisticated profiling technology to approximate sex and age.

But why the scare quotes? By any definition, what Scenetap does is anonymous. It is specifically designed and marketed to clubs and their patrons as a means for gathering demographic information and that information cannot be traced back to a specific individual because it uses no individual identifier such as a person's name (or cookie, but we'll get to that later). To go further and collect personally identifying information would require a real facial recognition system which would be very expensive, require a large investment in training and labor and probably wouldn't provide a sufficient return on investment (ROI) in a club/bar setting to make the effort worthwhile.

Then there's the conception of privacy in this passage.

This type of technology is already prevalent online, where customer preferences and habits are tracked — in order to recommend products or pages you may be interested in. As we cannot see the data being collated, it seems less of a privacy issue than knowing that cameras above are observing you — even though the information collected about your online activity is far more vast.

There's absolutely no equivalence between Scenetap and smartplanet.com. The image below shows that smartplanet.com places two cookies on a visitor's computer and runs seven programs in the background of which most users would be completely unaware: three for tracking the user; three for connecting to social media; and one to monitor the site's performance. One of the trackers, Crowd Science, even claims to be able to tell smartplanet.com about users' interests, preferences, lifestyles, attitudes, opinions and incomes.

Real world demographic analysis tools like Scenetap do no such thing. It's a dead certainty that smartplanet.com is collecting far more (and far more individualized) data, a fact that is acknowledged at the end of the quote.

Then there's the part where transparency and privacy are inversely related because "As we cannot see the data being collated, it seems less of a privacy issue than knowing that cameras above are observing you."

"Out of sight; out of mind" and "what you don't know can't hurt you" aren't theories of privacy one sees many people advancing these days. By this logic, bricks-and-mortar demographics analysis can attain smartplanet.com's level of respect for individual privacy by collecting vastly more information and using facial recognition technology to track individuals as long as they hide the cameras.

I don't want this post to come across as grousing about what web sites do. The folks at smartplanet.com are working hard to put food on their family just like the rest of us and people should understand that if they aren't paying, they aren't the customer; they're the product being sold. That's just the way it is. This is completely uncontroversial to those who operate in the online economy; but let a bricks-and-mortar organization deploy a tool that collects far less information and there's a tendency for those in the online world to come down with a collective case of the vapors. Physician, heal thyself.

See also:
Retail Marketing Technology Online and In Person
Transparency

Retail Marketing Technology Online and In Person

Not really biometrics related, but...
Software mines security footage to help business owners see what people do once they're inside the store (Technology Review)

"The huge success of online shopping and advertising—led by giants like Amazon and Google—is in no small part thanks to software that logs when you visit Web pages and what you click on. Startup Prism Skylabs offers brick-and-mortar businesses the equivalent—counting, logging, and tracking people in a store, coffee shop, or gym with software that works with video from security cameras."

Online retailers are able to free-ride on investments made by their brick-and-mortar competitors (see showrooming). They also have more powerful tools available to them for the purposes of analyzing detailed reports of user activity on retail websites. Why, the page I linked to for this story has fifteen programs that track your interaction with the linked page and TechnologyReview.com isn't even selling anything directly. The image to the left shows the list as compiled by the Ghostery add-on for Firefox.

If brick-and mortar retailers can't learn as much about customers in physical stores as web retailers know about user experiences, they must compensate in other ways or they're going to continue to struggle.

See also:
Target fights Amazon showrooming with plea for special product lines (ExtremeTech.com)

Face Recognition, Face Detection and Marketing

The New York Times takes a look at the marketing world's embrace of technologies that attempt to discern audience demographics to target advertising. It gets many important things right:
♦ Face detection to determine demographics is not face recognition for identification
♦ Anonymity is different than privacy
♦ Not all applications are created equal

The article, in general, is so solid that one could be forgiven for overlooking the clichéd yet obligatory Minority Report reference in in first sentence. [Was there even any face rec. in MR, at all?]

Face Recognition Makes the Leap From Sci-Fi (New York Times)

As SceneTap suggests, techniques like facial detection, which perceives human faces but does not identify specific individuals, and facial recognition, which does identify individuals, are poised to become the next big thing for personalized marketing and smart phones. That is great news for companies that want to tailor services to customers, and not so great news for people who cherish their privacy. The spread of such technology — essentially, the democratization of surveillance — may herald the end of anonymity.

And this technology is spreading. Immersive Labs, a company in Manhattan, has developed software for digital billboards using cameras to gauge the age range, sex and attention level of a passer-by. The smart signs, scheduled to roll out this month in Los Angeles, San Francisco and New York, deliver ads based on consumers’ demographics. In other words, the system is smart enough to display, say, a Gillette ad to a male passer-by rather than an ad for Tampax.

Burgeoning Facial Recognition: How come no pitchforks?

In the previous post, I linked an article about how facial recognition technology is really taking off.

M2SYS, made very subtle observation on the article:

In other words, "How come no pitchforks?"

He's right. New technology is often feared. In the case of biometrics, there have been those who have seen fit to feed that fear. Biometrics' association with security, and hence, authority would seem to put it right in the nexus between fear and fearmongering. Furthermore, the article's author chose examples of facial recognition applications that have gotten a lot of public attention and might be considered avant-garde to say the least. But so far, most public reaction seems to come somewhere between "cool" (Scene Tap), "that was dumb" (Facebook), or "about time" (London riots). Public comment has been rather moderate and even in tone. What gives?

I think there are several reasons why new facial recognition deployments are rarely met with scorn. Some are related to what it means deep-down to be human, and some are more practical.

♦ Except in a very narrow sense and among a very few cultures, a face is not, nor has it ever been, considered private. In the vast majority of times, places and cultures it has (like "name") always served as a proxy for an individual public identity. This fact is embedded in languages and useage all over the world. "He can't show his face around here anymore." "She really lost face." "They tried to save face."

♦ It's also probable that the vast majority of people intuitively understand the difference between privacy and anonymity. Privacy is the ability to keep things about yourself secret. Anonymity means "without a name". Privacy is a time-honored value that is nearly universal in city-building societies. Anonymity, freeing one from public scorn for one's non-private actions, has probably only been accessible to the masses in the relatively narrow space between the industrial revolution (that enabled rapid transit and the megalopolis) and the information age which seems to be making anonymity much harder.

♦ Marketers aren't stupid. When interractions aren't strictly voluntary, anonymity can act as a salve and biometrics can improve the efficacy and customer experience of marketing. Marketers know this and in a competitive marketplace the cost of mistakes is high. [After I've read it as well, and assuming the Phillip K. Dick story is similar to the movie, I'll write a detailed piece on how those who use Minority Report to say anything serious about biometrics profoundly misunderstand both Minority Report and biometrics. In the movie, the advertising posters were calling the name of John Anderton's eyeballs' former owner. Not only did this make the ads useless, it was also super-annoying to John Anderton. I would never go in a store where that happened and I doubt I'm alone.] The type of technology driving the variable ads is nothing like Minority Report. In fact, it's not really face recognition in the first place. It's really more demographic (in the marketing sense) recognition. It's non-individual (anonamous even) which is what makes demographics useful. And it's helpful. Even without biometrics, I see fewer ads for products I am not in the market for. Now if biometrics can just do something about all the perscription drug and trial lawyer commercials, I'm all for it.

♦ There have been no high-pofile victims of facial recognition gone awry — not that there won't be. It's easy to imagine that someone in a witness protection program or a high-value political defector might be tracked down and murdered in part using a facial recognition dragnet. Those people may want to swing by the CV Dazzle site for some face-rec-beating fashion tips. But it hasn't happened yet.

♦ Last (for now) but not least: After 9/11 there was a lot of overpromising an underdelivering about biometrics. A lot of money changed hands and a lot of people got egg on their faces. Still, they say "it's an ill wind that blows no good." The up-side of this bad history is that the public has had ten years to get used to the idea of what facial recognition applications are only just now begining to deliver. I think there's a sense among the public, at least on the security side, of: "I thought they were doing that already."

Undercover Feds Able to Easily Obtain Fraudulent e-Passports [USA]

The Center for Public Integrity

“The U.S. passport is the gold standard for identification. It certifies an individual’s identity and U.S. citizenship, and allows the passport holder to travel in and out of the United States and to foreign countries, obtain further identification documents, and set up bank accounts,” [Md. Sen. Benjamin] Cardin said. “We simply cannot issue U.S. passports in this country on the basis of fraudulent documents. There is too much at stake.”

Identity management is about people and trust.

Because the U.S. government, and the society from which its power is derived, is among the most trusted in the world, the identity documents that it issues are accepted with near universality.

The integrity of the process and the people involved in it is what confers legitimacy upon the document and its bearer and makes modern globalized travel (among other things) possible. This integrity of person and process, however, also raises the incentives to obtain a fraudulent U.S. passport precisely because it is so trusted.

In the linked article, the reason given for the State Department's issuance of five passports requested with fraudulent information is essentially that the system in place is under stress due to the  volume of identifications that it is requested to make with absolute certainty. Because the errors by the Bureau of Consular Affairs stem from over-stressing the passport issuing apparatus rather than corruption, it makes sense to try to improve how passports are issued rather than who issues them.

Technology can really help with this sort of challenge. Giving better tools to those trying to do the right thing makes the world a better place. Giving better tools to those who aren't trying to do the right thing can have the opposite effect.

In so many of the articles I read about new identity management technologies there seems to be an unstated premise that we in the industry are attempting to develop autonomous systems in order to control the behavior of ordinary people.

What is closer to the truth is that we are trying to help ordinary people make better decisions about who they can trust.

A lot about what makes living in the information age a wonderful thing is that we have the power to forge relationships (commercial, financial, romantic, etc.) with a quantity of individuals that is orders of magnitude larger than anything humans have ever experienced. This fact has forced people to come up with new tools to assess who is and who is not trustworthy.

Granted, these tools aren't always technological. The use of branding to confer trust is a very old technique indeed and it's actually pretty close to what the Senator is talking about in the quote above. The U.S. brand confers the trust that the identity document is accurate and authentic, but the U.S. can and does use technology in maintaining the value of its brand. If the brand wasn't worth protecting, no amount of technology could confer trust.

The linked article is worth reading in its entirety.

Japan Tests Gender-Aware Billboards

MarketingVox.com

Japan is testing a billboard that can tell the difference between male and female faces - and display appropriate ads accordingly. The system is running now in subway stations around Tokyo, CNET writes. A consortium of 11 railway companies launched a one-year pilot project to test the signs. Its aim, according to CNET, is to collect data on what sorts of people look at which ads at what times of day.

The CNET article linked above mentions that facial recognition is involved.

Tools developed for identity management can be applied to other challenges such as gender determination for the purposes of marketing in public places even though gender determination by humans is an extremely complicated process. Most facial recognition systems used for identification don't attempt to evaluate the gender of the individual. This is due to the fact that in a security context (logical and physical), the exact identity of the individual is more important to those who deploy theses systems than is gender. The users' tolerance for mistakes in the security context is low.

In a marketing context, the rules of the game change. A static billboard for a gender-specific product will appeal to some proportion of the population. Due to differences in age and gender that proportion is likely far short of 50%.

A system that tries to figure out age and gender doesn't have to be correct all the time in order to provide a positive return on investment (ROI) by increasing the percentage of impressions that have a chance of impacting customer behavior.

Moreover, the costs of the system making an incorrect judgment approach zero and the rewards for "tricking" the system are zero.

Even better, if the system is "tricked" by, for instance, a man with highlights and wearing lipstick, the system still works if you are advertising lipstick.

Sophisticated adopters don't let perfect be the enemy of the good.