Computer vision has been the talk of the retail industry recently—not because it’s new, but because the technology is finally advanced enough to make a powerful impact on loss prevention. The computer vision market in retail is expected to reach $11.4 billion by 2025, and 44 percent of retailers plan to leverage AI (which encompasses computer vision) in the next three years.
“If you look at the ability for computer vision to really perform today versus a couple of years ago, there’s been dramatic improvements, and the cost has gone down,” said Dustin Ares, LPC, director and general manager of video analytics, AI, and incubation at Sensormatic Solutions. “And more broadly, there are a number of use cases where retailers need help today with shopper behavior.”
Computer vision works to automate tasks and derives meaningful information from video footage in real-time. Sensormatic Solutions’ innovative computer vision technology delivers retail operational insights based on best-in-class deep learning AI models. These solutions were created in partnership with Intel and optimized for retail using Sensormatic’s vast retail experience to inform its AI development process.
“[Computer vision] can help the customer in a few different ways,” Ares explained. “We think of our solutions from a shopper perspective, and the first questions retailers are trying to answer are ‘Do I have the right product in the right place at the right time?’ and ‘Do I have the right people in the right place at the right time?’ So [our computer vision solution ensures] you have the right people in the right place, where they can answer questions and carry out their tasks to make the shopping experience better. And then we make sure products are in-stock, in the right spot on the shelf.”
Diving deeper, computer vision can determine whether promotions should be created to drive different focal points within the store, or whether certain areas create risks for customer injury or theft activity.
“When we apply computer vision to the shopper journey, we can determine the coordinates shoppers are moving in within the store, and over time we can use AI to determine what areas of the store have a lot of visitors, and even the dwell time in those areas,” Ares said. “And you can see dead zones that become focal points for change. After you make changes based off of those trends, we can compare pre- and post-change and tie it back to sales, because at the end of the day you’re trying to figure out how to drive more sales.”
Computer vision can also tell if someone takes too many items off of a shelf at the same time, as that could be a sign of theft. The technology can then notify employees in real-time, so they can intervene with good customer service.
The beauty of computer vision is that it can make the security technologies you already have even more powerful.
“We think about security as being in layers, and computer vision is a layer that augments other security systems within a store,” Ares said. “Think about EAS or RFID as one method to detect an item leaving the store without being paid for, and computer vision can validate that it is a likely theft event. Then, in the future you will understand whether that same person victimized you previously, and you can build a case history. There are privacy concerns and approaches, but ideally in the future you will be able to determine before a person enters a store, whether there is a risk being presented.”
As computer vision becomes more advanced, Sensormatic is careful to keep privacy top-of-mind.
“In Europe, the US, and other countries there is a concern that this technology could be used to track specific people,” Ares explained. “What we do is vet our AI models to ensure we are not violating any regulations. A key foundation of our design is using a method called privacy by design that ensures our AI models are not storing or maintaining any privately controlled information about a person—no facial images or anything like that—unless the retailer is requesting that information specifically for a purpose like a known offender registration.”
Looking to the future, Ares is excited about how computer vision will continue to improve, and how Sensormatic will be able to serve retailers even better.
“[In the next five to ten years] I see computer vision as a key layer in the evaluation of what the shopping experience should and shouldn’t be,” Ares said. “It should be detecting attributes about shoppers, what categories they fall into, what they’re interested in, and the marketing should follow that. If you’re walking up to a display, it might change based on who you are and what you might be interested in. On the converse of that, if you were a known shoplifter, we would add friction at every point in the process to deter you.”
Learn more about how computer vision can transform your loss prevention program at sensormatic.com/computer-vision.