NTechLab’s FindFace Facial Recognition Technology Has Applications In Retail, Public Safety

FindFace started as a futuristic social technology for identifying strangers by scanning their faces with a smartphone camera. Two years later, the facial recognition technology is the best in the world (yep, even better than Google’s) and is being used for public safety, law enforcement and fraud prevention through cybersecurity.

Of course, facial recognition has driven significant public controversy over the erosion of personal privacy and anonymity. People also worry that their personal biometric data could be stolen and used for nefarious purposes.

For Mikhail Ivanov, CEO of the Russian machine learning and artificial intelligence company NTechLab (which created FindFace), that’s no reason not to innovate. It just means that society needs to start having a conversation about when and where to welcome the new smart technology.

“We have to regulate how to use it in the future. It already exists,” Ivanov said. “Everything can be useful for good things and for bad things. I think there are lots of cases where facial recognition can be used for good things to make our lives safer and more comfortable.”

For instance, Ivanov said, the government already has personal data about its citizens. Biometrics and facial recognition are just another piece of information it can use to do its job and protect people in a public safety and law enforcement capacity. Police already use it to match driver’s licenses.

Cameras are already everywhere, Ivanov said. The amount of information they collect in a single day is too much for a human to sort through. FindFace can spot criminals based on information in the police database, making it possible to actually do something with all that footage.

That’s a pretty straightforward application and has, in many instances, already been tapped. But there’s a lot of untapped potential in other verticals, particularly retail.

Retail needs some help. E-commerce has changed the landscape, making it harder and harder for brick-and-mortar stores to succeed. Shopping online is cheaper and easier. And so is selling online: Based on activity, targeted offers and ads can be pitched directly at the consumer. Brick and mortar doesn’t have that luxury. But with biometrics like FindFace, it could.

“In offline retail, there’s no key to getting access to your previous purchases before the cashier desk,” Ivanov explained. That means that employees may or may not understand what a shopper is looking for and that person’s buying history. If the guest leaves without buying, the store will never know why.

With smart technology like FindFace, the face becomes that key. It unlocks data about the shopper, routes it to a salesperson’s tablet and facilitates an efficient and satisfying shopping experience.

Imagine walking into a store and immediately being greeted by an employee who knows exactly what you’re shopping for? They don’t waste time with redundant questions; they already know what brand you bought last time and how long ago that was. They can see the online research you conducted before setting foot in the retail location.

Facial recognition could also trigger a VIP experience for loyal customers, giving them more reasons to come back and shop again.

Meanwhile, as it’s providing this enhanced experience for the guest, the same biometrics platform can gather rich statistics about demographics, including age, gender and whether customers have shopped there before. That can offer critical insights for marketers: If they’re trying to attract people over the age of 35 but all of their customers are much younger or older, it may be time to change strategies.

Of course, since it’s already integrated with the surveillance cameras, FindFace can also be used to identify shoplifters — sometimes even before they steal, if they are repeat offenders whose face has been flagged — to protect the business and decrease losses.

In other verticals, FindFace can avert fraud through cybersecurity in financial services by requiring biometric inputs to access personal data and accounts. Car insurance companies can use it to confirm that the person who is allowed to drive a car was, in fact, the person driving it at a particular time and place.

NTechLabs engineers employ deep learning strategies to teach the FindFace neural network the differences between faces.

Like any machine learning, this takes a lot of trial and error. In the first stage, engineers show pictures to the artificial intelligence and ask it simply to identify which ones are faces. They correct errors and add layers of complexity until the AI is able to differentiate between a whole gamut of faces: Asian, European, African. And then it learns to look past occlusions like hats, glasses or face masks.

When FindFace was finally smart enough to go public, it launched in 2016 as a tool for young urbanites to identify one another in public and connect later on social media, rather than approach someone at random on the train.

Russian founders Alexander Kabakov and Artem Kuharenko soon learned that there were myriad applications for the technology and set about creating a version that could be used at a higher level for the greater public good. Thus, FindFace Pro was conceived.

FindFace can now recognize faces with 73.3 percent accuracy, which is 5 percent better than Google’s facial recognition technology. But that doesn’t mean people are lining up to use it. Because the technology is so new, people don’t yet understand how it can be implemented or how it could help them. Plus, there’s still that pesky social discomfort about violation of personal privacy.

Ivanov said NTechLab is approaching the market with proposals to integrate FindFace at their business, not the other way around. So far the technology has been licensed by fewer than 100 companies. But there are several pilots currently in progress, and Ivanov expects to deploy FindFace in markets across Russia, Europe, the U.S. and Asia in the coming years.

“We are … number one in facial recognition technology around the world,” Ivanov said. “That’s the engine. But you have to take the engine and apply it to this or that use case or scenario.”