For a minute, envision a movie or TV show featuring a law enforcement agency looking for the bad guy who is trying to flee the country. Cut to a computer screen using facial recognition technology to scan every person picked up by a CCTV Surveillance camera at the airport (CCTV stands for closed-circuit television). Within minutes the missing person is identified by the computer and apprehended. Is this speed of identification really possibly or still in the realm of science fiction? I decided to do a little research to learn the answer.
While not yet ubiquitous, facial recognition technology is being used in the real world for a variety of applications. Facebook uses facial recognition technology to tag people in the photos you upload, and many digital photography applications can recognize, sort and identify the people in your photos. If you use an iPhone X, Galaxy Note 9 or an LG G7 mobile phone, you can unlocked it by holding the phone up to your face. Apple calls their system Face ID, and according to the Apple website, “Face ID uses advanced machine learning to recognize changes in your appearance. Wear a hat. Put on glasses. It even works with many types of sunglasses.” To train a device, the user takes multiple photos of their face, and software builds a facial map to use for identification.
On the other side of the world, Australia has been in the forefront of facial recognition technology since 2007 with the system-wide introduction of SmartGates border control. The system is used for holders of ePassports entering the country at its international airports. ePassports, available in Australia and a number of other countries, including the United States and Canada, include an embedded, biometric chip containing a photograph and personal data. Upon entry in Australia, SmartGates uses facial recognition technology to verify travelers identities by comparing a facial scan to the data stored in the passport, in addition to information housed in international immigration databases.
In 2017 it was estimated that the police in London monitor over 51,000 cameras, and that the average Londoner was captured on camera about 300 times a day. Closer to home, New York has deployed approximately 6,000 CCTV cameras, as well as using license plate readers that capture photos of all cars traveling in and out of the city. The surveillance system in New York is one of the most sophisticated in the United States. The city is working with Microsoft to employ smart cameras using Artificial Intelligence to flag suspicious behaviors, such as vehicles that repeatedly drive around a singular location and unattended packages in public places. Chicago police command center’s continuously monitor approximately 10,000 cameras 24/7. These numbers do not include the thousands of private surveillance cameras installed in many of these cities. What is not widely known is how many of these CCTV system employ facial recognition as part of their regular surveillance procedures.
Many retail stores, as well as public sports and entertainment venues, employ cameras to take photographs of patrons, and compare the photos with databases of known criminals. However, you don’t have to be a criminal to have your photo included in a database of faces. According to reports by several news organizations, including both NBC and CBS news, the faces of most law abiding, adult Americans are part of statewide facial recognition databases. The photos are obtained from driver’s license ID photos, social media, and various government IDs. It is up to the states to decide to make their databases available to law enforcement, and presently sixteen states allow FBI access.
As with most technologies that use AI, this type of surveillance and identification is in its early stages, and privacy, misuse, and abuse are issues that must be addressed. Recently, facial recognition software designed by Amazon, Microsoft, IBM and a Chinese firm Megvii, have come under fire for problems of mis-identification as well as gender and racial bias. Last year, when the ACLU conducted a test of the Amazon software by scanning photos of members of Congress, if found that the software falsely matched twenty-eight of the photos with police mugshots. Citing privacy concerns, state and local governments are beginning to pass laws and issue guidelines to regulate the use of facial recognition software.
It is pretty obvious that facial recognition is going to become a dominant form of identification in the future. What today’s column hasn’t answered is the question posed in the first paragraph. Facial recognition technology does work, abet with many inaccuracies, but nowhere could I find information about the time required to match a known individual against a sea of faces appearing on real-time CCTV cameras. Thus it will probably be a few years before the scenes we see in the movies become everyday occurrences in real-life law enforcement. Look forward to that information in my 2030 column (lol).
No comments:
Post a Comment