The business of facial recognition software
What does the largest search engine in the world have in common with the largest casino in the world? Surprisingly, both of these corporate organizations, Google (headquartered in the United States) and The Venetian Macao (headquartered in China) make extensive use of facial recognition software. Not surprisingly, this software is also utilized by many governmental agencies including The United States Department of State and National Security Agency (NSA).
Historically, facial recognition software -- produced using algorithms which integrate mathematical concepts with names such as the Hidden Markov model -- have had a mixed reputation because common environmental factors -- like light or lack thereof -- can have a great impact on the performance of the system. So can a new hairstyle. Or, for that matter, a new pair of glasses. Of course this hasn’t stopped such algorithms from being integrated into hardware and software products by their designers as a feature that provides strategic differentiation against competitors.
Google, as mentioned earlier, includes facial recognition capabilities in its popular Picasa photography product. This specific capability, as a secondary feature used in tandem with the other primary photography editing and posting features, provides a better experience for users and helps customers efficiently index large set of photos by grouping them based on recognized faces. To contrast, the Venetian Macao, which has over 12,000 employees in the casino daily, uses facial recognition software to prevent impersonators disguised as employees from entering the casino by using the identification of another individual.
The software, depending on the needs of the purchasing organization, can be used in a seamless fashion -- as part of a more comprehensive product like Picasa in which the facial recognition algorithms are embedded -- or as a standalone security product. This is how Bioscrypt’s VisionAccess 3D Face Reader (now owned by L-1 Identity Solutions) is used by the Venetian Macao.
One popular platform for facial recognition software is Face.com. The company is known for producing a software that scans users' Facebook photos and identifies which of their friends appear, adding a tag for each one. As of February 2011, the software had 'found' 18 billion faces of friends in users' photos.
Recently, Face.com rolled out its latest update to its API, which claims that it can determine a minimum, maximum and estimated age for each face it scans. The technology was developed to be integrated into apps or sites that contain age-sensitive material to prevent underage individuals from accessing it. It also has other obvious uses, such as verifying age in liquor stores and nightclubs, or for the purchase of cigarettes. The company says that the software is now 30 percent more accurate with the newest update.
Prior to recent news advocating the benefits of such technology, it wasn’t very long ago that facial recognition platforms had a mixed reputation, specifically around their ability to be both accurate and dependable. HP, a leader in the personal computing space, learned about the downside of facial recognition software in 2009 when they decided to include the technology in their laptops' web cameras as a security feature. Unfortunately, these inexpensive web cameras did not perform well in low light conditions, thus system owners found themselves locked out of their own computers.
Outside of technical feasibility, these deficiencies are a serious issue for hardware manufactures. Integrating third party technologies which sound impressive, but then underperform, can be detrimental to a company’s brand and, more importantly, its sales. Back to HP, the story actually gets worse. It turns out their facial recognition software had serious problems recognizing users with dark skin, so the computer maker was forced into a defensive posture when their product line was branded 'racist' as it only acknowledged Caucasians.
Talk about a public relations debacle.
.jpg)
However, despite past failures, other companies are also testing the waters with facial recognition in their own products. Google is using the technology in its Android 4.0 and Apple applied for a facial recognition patent in 2011.
Microsoft has also made great strides in improving facial recognition software by incorporating new capabilities which address traditional deficiencies of the past. The Kinect, used for gesture recognition in Microsoft’s Xbox 360 gaming system, also has facial recognition capabilities -- instead of depending on natural light, however, it makes use of infrared so that low lighting is not an issue. The Kinect also uses three-dimensional mapping which, unlike the two-dimensional systems of the past, ensures a greater degree of accuracy because of the introduction of additional measurement variables.
The intelligence behind facial recognition software has been growing at an exponential rate with the introduction of new variables that improve the efficiency of the underlying algorithms. This in turn creates windows of opportunity that can either be overlooked or capitalized upon. Regardless, proper due diligence such as user testing is considered not only a best practice but a necessity when integrating third party technologies into a company’s flagship products. Face.com, for instance, made a wise choice to innovate and improve their products by introducing a crowdsourcing component to increase the technical reliability of their solution.
Technological innovation, in our modern age of disruption, is just as much about good offense as it is good defense. A decade ago, Google and The Department of State probably didn't use many of the same tools. Even though the organizations have very little in common, with the changes in both technology and the way we use it, they were each able to re-position the same technology for their own purposes, showing that no matter how complex technology gets, we will always find a use for it (or several).