Facial recognition has come a long way from being a dream to being used in a number of everyday applications. The latest development in the technology is Chinese cops making use of it in special sunglasses. The glasses have a special camera attached for facial recognition as reported by The Wall Street Journal.
The Chinese system also involves a connected mobile device in addition to the glasses which contains offline face data. This allows the system to work more quickly. They have already taken seven people associated with crimes into custody using the facial recognition glasses.
The system follows the same procedure used by other facial recognition software. This is a two-step process as explained by David Alexander Forsyth, an artificial intelligence expert and chair of computer science at the University of Illinois at Urbana-Champaign.
The first step is to figure out where the faces are in the image. The system looks for a window-like section of the image that has someone’s face or facial expressions on it and ignores the other stuff. The second step is to match that face to any of the faces in its database which is not as easy as it sounds as explained by Forsyth. “Turns out, that’s a harder problem. People tend to look like each other.”
The system does not look at the images the way a human would and looks at them as a representation of data. “That representation has to emphasize things that make people look different from each other,” says Forsyth. These include details involving the shape of features like lips, noses, and eyes. The representation also needs to make sure that it is unaffected by variables such as light falling on someone’s face. The software then examines this data to see if there is a match with a face it has in the database.
“The last 10 years or so have seen amazing advances and changes in classifier technologies,” he adds. “The procedure of building that representation of the image has become extremely sophisticated and very effective.” AI systems need a lot of data in order to perfect their job and facial recognition technology is the same. Forsyth says, “Right now, the best way we know, by a long, long way, is to have an immense number of pictures of faces.”
Besides the numerous advancements in technology, it is still a difficult task and “The consequence for a mixup can be truly terrible.” The software can have false positives when it thinks it has a positive match to a person of interest whereas, in reality, it is not the same person.
The technology is somewhat different from the facial recognition technology in phones. The phone only needs to recognize your face which you feed into the system yourself. It doesn’t have to pick out your face from millions of files on the database. Even if it fails to recognize your face, you can simply use the passcode to unlock your phone and there are no serious consequences.
However, using this technology to scan millions of faces in crowds and settings like airports or train stations, things get tricky. If a false match occurs, it is not only a cause for embarrassment for the law enforcement agencies but also wastes valuable time of the travellers. The system still has room for improvement and it will need millions and millions of photos before it can perfect its facial recognition capabilities.