Imagine being able to identify a shoplifter or criminal as soon as they walk onto your premises. Imagine an era in which criminals cannot run and cannot hide, but are spotted and their plans thwarted before they can even think twice about doing something illegal. That era is upon us now.
Facial recognition technology is changing the face of video surveillance and loss prevention as we know it. Until recently, it was only possible to identify somebody as a criminal after they had committed a crime. While this is great for preventing future crime, it does nothing to help stop crime before it occurs. Well, NTechLab’s FindFace algorithm is about to change all that. This revolutionary technology gives us the capability to not only determine who someone is, but it also allows us to know where they’ve been, where they’re going, whether or not they have an outstanding warrant or unpaid parking ticket and if they have been caught shoplifting before, amongst other extraordinary capabilities.
In the past, retail security camera systems have been able to deter crime, but they haven’t been able to eradicate it completely. That may be about to change. With facial recognition technology and advanced CCTV surveillance systems working together to monitor retail locations and other businesses, criminal behavior has become a science that we are just now finally beginning to track and understand. And the results are phenomenal.
As more and more retail locations use facial recognition to identify criminals as soon as they enter the premises, criminal patterns are becoming more and more obvious. According to one user, facial recognition has allowed them to anticipate when a shoplifter will return to their store. They know that 26 percent of people they detain will be back within one month—more specifically, criminals return within an average of 13 days. This is behavior we have never had an opportunity to analyze before, and it’s changing the video surveillance game entirely.
Moreover, this technology has helped retailers track which times of day are most active for shoplifters. Having a firm understanding of which individual thieves enter which stores and the distribution of peak shoplifter activity over the course of a single day, month or year can be extremely powerful to store managers, loss prevention teams and law enforcement alike. Now, instead of being reactive to shoplifting and other crime, stores can be proactive and thwart criminal activity before it even occurs.
Let’s say that Jane Doe is caught shoplifting at a Wally World branch in East Texas. Her photograph is taken upon detention, she is given a barring notice and her information is entered into the Wally World database. Three weeks later she walks into the same branch. The video surveillance system captures her image, sends it to the database and within seconds, has compared it to the system’s enrollees and identified her as a match. An alert is sent to the store’s loss prevention team, and a member greets Jane at the door, only to show her the way out. Within minutes the team has successfully helped a shoplifter leave the store empty handed.
A database is populated with an enrollee’s facial biometric information. This can include the length of a known shoplifter’s nose, the distance between their eyes and a mole on their left eyebrow. Amazingly, each shoplifter’s photo consists of tens of thousands to millions of these identifying parameters, making it difficult for an innocent shopper to be mistaken for an enrollee.
On an everyday basis, the system’s security cameras are connected to the database and the detection algorithms. The cameras then do most of the work. They scan and analyze each face that enters the store—even faces that enter in crowds—and the system intelligently selects the best possible angle and lighting condition from the video footage to ensure the best possible match. Each shopper’s image is then analyzed and scored against the database’s profiles. If no match is found, the shopper’s photo is deleted from the system. If a match is found, an alert—along with a photograph—is sent to a designated store associate’s cell phone, who is then charged with finding the shoplifter and escorting them out.
If a person is caught shoplifting on camera but was not caught by a loss prevention associate and made to sign a barring notice, the store can use a good photo from the CCTV system’s footage to input in the database. In this instance, the store clerk will still be alerted, but they will not approach the individual; rather, they will observe them carefully until they leave the store.
Finally, alerts can be set for violent criminals. Upon the approach of a known violent offender, the store clerk is notified to call the police immediately instead of approaching the criminal themselves.
As of right now, there are a few issues surrounding facial recognition technology. For one, whether or not a face is a match is not as simple as looking at a photograph. For all intents and purposes, an individual may look like a previous shoplifter, but if a biometrics analysis does not turn up a match, the store is not allowed to take proactive measures. However, the system works based on how probable it is that a face on the security camera matches one in the database, and each store can set its own “probability threshold.” This, in itself, is a limitation, as if they set it too low, they may end up stopping every John, Dick and Nancy that enters the store; however, if they set it too high, they might miss a known shoplifter if the camera’s angle isn’t just right.
Aside from that single limitation (which is able to be overridden by the store manager), another issue presented by facial recognition technology isn’t so much a limitation as it is a privacy concern. On the one hand, facial recognition as it is being used today poses very little privacy threat: images are captured, analyzed and then deleted so long as a match doesn’t come up. If an individual chooses to commit a crime, they subsequently forfeit their right to. In that regard, privacy is not an issue. But it could become one.
As more stores realize the marketing potential posed by not just facial recognition, but also by the identifying information that pops up after a thorough analysis, they may decide that they can use that information to create the “ideal shopper persona.” Some might even go so far as to sell that information. However, as of right now, this is a very slim possibility, and so long as retailers continue to use the technology for what it was made for—crime prevention—there is no need to worry. If, in the future, information does become abused, there is no doubt that regulations will be put in place to protect consumer privacy.
In fact, establishing standards up front may prove to be the saving grace that this technology needs to convince skeptics that this technology is about protecting them, and not about exploiting their information. Some possible standards to consider are enrollment and unenrollment procedures, data storage and protection, data sharing, data ownership and transparency, among other things. Stores can also start implementing signs that read, “Shoplifters Beware: we reserve the right to use biometric ID to identify you for the purposes of enforcing a barring notice.”
Though there will certainly be challenges posed by facial recognition technology, as it becomes more widely accepted and adopted, those challenges will (hopefully) be properly addressed. This technology is not an invasion of privacy, but rather a way for stores to finally put to use information that they’ve spent precious billable hours cataloging. Now, instead of information on shoplifters being left in boxes in a warehouse serving no real purpose, that information can be digitally stored and used when it really matters—right before a crime is about to happen.
Boston’s dense population results in a high crime rate. According to statistics, Boston has a total of 21,503 crime incidents on average per year, with a crime rate of 32.23 crimes committed per 1,000 resident. Of the 21,503 total, 16,732 of which are property crimes. This makes Boston a risky place for property owners whether it’s for residential or business use, as the potential for burglary, illegal settlement, and vandalism is high. Mammoth Security security cameras are compatible with facial recognition technology and can be used to capture the high resolution footage you need to tap into your database and identify a criminal. If you own a business in Boston, don’t take any chances, and speak with our surveillance system installation Boston team about how we can protect you.
High Tech Security Systems to be Installed on Busy Transit Systems in Boston – new post in the blog!