- Digital Democracy makes you ‘a fly on the wall’ at the state house
- Facial recognition body cameras turn police officers into mobile surveillance systems
- With facial recognition software, diagnosing genetic disease is as quick as taking a photo
- Fighting thieves with new-age mug shots
- No more hide-and-seek
Remember that moment when you saw someone on the street and were sure you’d seen their face before? You searched your memory and then suddenly you got it — you remembered her from your trip to the grocery store the previous week. The biological process of facial recognition led you to this conclusion. Each person has distinctive facial features: their eye shape, or maybe a big forehead or a pointy nose, and we use these features to differentiate between people at a glance.
Nodal points – those features we don’t immediately recognise
However, there are some delicate characteristics that we’re not able to recognise immediately, things like the distance between a person’s eyes, the depth of their eye sockets, or the length of their jaw. These are known as nodal points, and researchers suggest that each face has approximately eighty of them. We’re not able to process those features because we look at the face as a whole, but they’re a distinctive marker of identity. ”Facial recognition software is already quite accurate in measuring unchanging and unique ratios between facial features that identify you as you,” Jan Chipchase, a facial recognition researcher insists. “It’s like a fingerprint.”
In the last few years, in part due to the explosion of AI, facial recognition systems have developed rapidly. You’ve probably seen at least one episode of CSI, where investigators are able to catch a criminal and solve a crime based on high-tech facial recognition tech. In reality, it’s much more difficult than that, and though facial recognition technology has progressed, it’s still far from what we see on TV shows like CSI.
Teaching a computer how to differentiate between mugshots
Facial recognition has been an ongoing project for decades. The pioneers in automated facial recognition were members of a team led by the American computer scientist and one of the founders of AI, Woodrow Wilson Bledsoe. Bledsoe, along with his colleagues Helen Chan Wolf and Charles Bisson, worked on using computers to recognise facial features. Throughout 1964 and 1965, they used a database consisting of mugshots to teach a computer to differentiate between faces. This process wasn’t completely computer-based since humans determined the coordinates on each face from the photograph, and then put those coordinates into the computer to process them. But by doing so, they established the nodal points in the computer’s memory, allowing it to establish and remember such things as the width of the mouth and eyes and distance from pupil to pupil. Using this method of facial recognition, the computer could process about forty photos an hour. This experiment was largely secret, and in the following years, similar tests were conducted using the Bledsoe process. Some of these demonstrated fascinating results; computers, for instance, were able to surpass humans when given facial recognition tasks because of their ability to see the small differences we lose in the big picture. This early work led to refinements in the process, and now, facial recognition uses two main, and various other approaches.
Geometrical, pictorial, 3D and thermal imaging recognition
In the first approach, computers can use geometrical recognition to differentiate faces from one another. This technique uses distances between facial features like the eyes, nose, and mouth to ‘map’ a face. The second is pictorial recognition in which the patterns of facial features across the entire face are recognised in relation to one another. Using one or both in combination, modern facial recognition is sensitive to even the smallest details that make each person’s face unique. This can allow a computer to recognise a face even when its expressions change.
A milestone in facial recognition technology was the introduction of three-dimensional facial recognition. A problem plaguing facial recognition is its inability to function in less than ideal circumstances like poor lighting or partially revealing angles. As Ralph Gross, a researcher from the Carnegie Mellon Robotics Institute, explains, ”Face recognition has been getting pretty good at full frontal faces and 20 degrees off, but as soon as you go towards profile, there’ve been problems.” But by using 3D sensors, computer systems can better determine facial shape and then identify different features, rotating the image, for example, or recognising a face in profile by extrapolating features from the ones it can see. Adopting this 3D approach, developers are working on further refinements to facial recognition. By using three cameras pointed at different angles toward a subject’s face, precise facial detection and recognition is possible, helping to mitigate the weaknesses of facial recognition technology.
Another new approach to improving facial recognition involves thermal imaging. In situations where a subject’s head is covered with a hat or his eyes are covered with glasses, thermal cameras come in handy. This type of facial recognition system detects only the heat given off by the head, ignoring accessories that fool the eye. Nevertheless, as in all facial recognition systems, the tech is only as good as the database, and thermal images for facial recognition are in short supply. Another promising innovation is skin texture analysis. This technique detects specific spots and lines on the face, converting them into a mathematical map. Think of this as a variation on the facial geometry approach that uses a different set of features.
Facial recognition tech in action
Facial recognition is still in its technological infancy, but in a relation to other biometric techniques of identification, it’s the only one that does not need subject’s cooperation. As a result, it’s a useful tool for securing public places have facial recognition systems installed without people being aware of them at all. From its beginnings in the 1960s until now, facial recognition technology has come a long way. And governments, police departments, and security agencies are not the only ones who incorporated this technology into their work—soon, your face might replace your pin at the ATM! And though the ubiquity of facial recognition technology raises privacy concerns that should be taken seriously, there are already numerous examples of this tech in action.
1. Digital Democracy makes you ‘a fly on the wall’ at the state house
Sam Blakeslee, the founding director of the Institute for Advanced Technology & Public Policy at California Polytech, recognised the importance of government transparency, and importance of unveiling the ways the elected officials operate. To achieve greater government accountability, teamed with students, he began working on the Digital Democracy platform, funded by the Laura and John Arnold Foundation and the Rita Allen Foundation.
The Digital Democracy platform relies on bots that record a lawmaker’s every word, and use sophisticated facial-recognition software so we know who’s talking. To avoid the platform being deemed a ‘data dump,’ the transcripts will be sorted by an AI tool called ClaimBuster that automatically identifies ‘assertions of fact,’ and transfers the claims to PolitiFact, America’s Pulitzer Prize-winning fact-checking website.
2. Real-time facial recognition body cameras turn police officers into mobile
surveillance systems
Citizen security has always been a priority, and with the latest face-recognition technology, it will be possible to advance the existing surveillance system. Namely, facial recognition startup NtechLab, founded in 2015 by Artem Kuharenko, managed to create algorithms ‘as intelligent as humans and as efficient as machines.’ NTechLab’s FindFace, once integrated with police body cameras, will enable faster identification of a person, and can help the police determine someone’s whereabouts or identify suspects.
However, not everybody welcomes the new facial recognition technology. For instance, Jake Laperruque, a fellow at the Constitution Project, expressed his concern regarding invasion of privacy. “These cameras are small, hard to notice, and all over the place. That’s a pretty lethal combination for privacy unless we have reasonable rules on how they can be used together.” Nevertheless, law enforcement is embracing this tech. According to a study of face recognition technology published by Alvaro Bedoya, executive director of the Centre on Privacy and Technology at Georgetown Law, at least five U.S. police departments have either already got the new real-time face recognition tech, or considering the option of implementing it.
3. With facial-recognition software, diagnosing genetic disease will be as quick
as taking a photo
The Boston-based startup, FDNA, has developed a program Face2Gene that helps physicians discover genetic syndromes that leave tell-tale signs on patients’ faces. Researchers insist that the software that scans a patient’s face is not a diagnostic tool, and should instead be used as a ‘browser for genetic diseases,’ serving as a search and reference informational tool. What’s so great about the program is that photographs of the patients worldwide can be uploaded on Face2Face, expanding the database.
Researchers with the National Human Genome Research Institute (NHGRI), part of the National Institutes of Health (NIH), have successfully used facial recognition software to diagnose a rare, genetic disease known as DiGeorge syndrome and velocardiofacial syndrome. Maximilian Muenke, M.D., atlas co-creator and chief of NHGRI’s Medical Genetics Branch, emphasised the importance of the facial recognition software for early diagnosis. “Early diagnoses mean early treatment along with the potential for reducing pain and suffering experienced by these children and their families.”
4. Facial-recognition tech fights thieves: new-age mug shot
Facial-recognition technology has proved efficient in taking on thieves, and Beijing might be a perfect example how state-of-the-art technology is used to tackle one of the city’s biggest issues – toilet paper thievery. The capital city’s Temple of Heaven Park introduced facial recognition technology, and the process is rather simple. A machine first scans a visitor’s face via a built-in camera, and when it confirms that you’re not the same person who requested toilet paper a couple of minutes earlier, it will provide you a strip of paper. If a person wants more paper, she must wait for nine minutes!
Face recognition tech: no more hide-and-seek
It took decades for facial recognition technology to be useful. Though the first experiments were promising, paving a path for future research, it took decades of hard work to deliver face recognition systems that are useful for security. But today, high-tech gadgets that rely on this technology have become widely used in numerous sectors, and they enable us to more easily catch thieves, secure airports, or provide patients with quick and easy diagnoses.