As the holidays approach, loss prevention becomes an even bigger concern for retailers than usual. That’s because shrink increases by 15 percent when the holiday season arrives, and the holiday season as a whole can account for as much as 37 percent of a retailer’s total annual shrink. Shrink losses may be even higher this season, considering there’s a clear correlation between rising inflation and increased theft.
With this in mind, retailers may wish to consider using every tool at their disposal to deter opportunistic thieves and organized retail crime (ORC) organizations. One technology in particular that can deliver impact for retailers of all types is computer vision.
Let’s unpack what computer vision is, how it works, and all the benefits it can offer retailers this holiday season—and beyond.
What Is Computer Vision?
By now, you’ve probably heard of computer vision in some form or another, but what is it, exactly? IBM offers a concise definition:
Computer vision is a field of artificial intelligence (AI) that enables computers and systems to derive meaningful information from digital images, videos, and other visual inputs—and take actions or make recommendations based on that information. If AI enables computers to think, computer vision enables them to see, observe, and understand.
In short, computer vision serves as the “eyes” of a machine learning system. This empowers connected solutions to analyze video data in near real-time. For retailers, this means having a dedicated set of eyes to help watch shelves, aisles, checkouts, entrances, and even parking lots, and monitoring for a variety of use cases. The results of this monitoring can come in many forms, including (but certainly not limited to):
- Creating heat maps based on actual consumer traffic to optimize the floor layout.
- Measuring store traffic, pass-by traffic, display interactions, dwell times, and more.
- Measuring audience response to in-store advertisements.
- Measuring inventory levels.
- Recommending stocking and auditing planogram compliance.
That’s a lot of upside from a single technology, but perhaps the biggest upside of all comes from what computer vision delivers for loss prevention teams.
The Future of Loss Prevention
With the retail labor shortage showing no sign of slowing, today’s retailers are continuing to ask stores to do more with less. That means fewer associates on the floor—and fewer eyes watching for dishonest behavior in the store and on the security monitors. But with computer vision, those are eyes retailers can allocate to other tasks. After all, from a loss prevention and security standpoint, computer vision can monitor for:
- Shelf sweeps.
- Associates removing tags without scanning items.
- People loitering on premises after hours.
- Suspicious and aggressive behavior.
- Arrival of large groups of individuals, such as flash mob or ORC scenarios.
- Person of interest detection.
When a computer vision solution detects any of these, it can send automated alerts to store personnel and loss prevention teams who are able to take action immediately, rather than reacting after the fact (and that’s if an associate witnessed the incident or anomaly at all). This is a huge win for retailers, yet it’s far from computer vision’s full potential. “Traditionally, computer vision has been used for object detection,” said Dustin Ares, general manager of video analytics, AI, and incubation at Sensormatic Solutions. “That includes everything from people counting to mask detection and more. These use cases are great and have a lot of value for retailers, but the biggest contribution a computer vision solution can deliver comes from recognizing behaviors, not just objects.”
Ares noted that internal theft is also at play, with more than 50 percent of retailers surveyed in the National Retail Federation’s 2021 National Retail Security Survey reporting that combatting internal theft has become more of a priority for their organization. This includes things like dishonest associates taking something from the cash drawer and putting it in their pocket, propping the cash drawer open with their hip in anticipation of a later theft, or even conducting a refund without a customer present. It also includes shopper behavior, like:
- Scan avoidance at self-checkout.
- Entering a store with a tool or device intended to deceive or destroy security equipment.
- Concealing merchandise in the aisle or staging merchandise at an emergency exit for them to take later.
Each of these behaviors, Ares noted, can be detected by a computer vision solution once it’s identified what “regular” shopper behavior looks like versus “anomalous” behavior.
“We can implement a computer vision solution that monitors for—and helps detect—virtually any anomalous behavior a retailer is interested in identifying, as long as that behavior takes place in front of a camera,” he added.
Major Return on Relatively Little Investment
A technology this powerful—with the ability to deliver benefits to loss prevention, customer experience, labor optimization, and across the entirety of the retail enterprise—must come with a hefty price tag and an intense implementation process, right?
Not so, according to Milton Navarro, video solutions manager at Sensormatic Solutions. “Retailers are often shocked at how cost effective and easy to implement computer vision can be,” Navarro said. “Especially if they already have robust camera infrastructure—and many major retailers do.”
Navarro pointed out that some computer vision solutions can work with retailers’ existing camera infrastructure, meaning there’s relatively little hardware to buy. That’s certainly the case for Sensormatic Solutions’ computer vision solutions. Unfortunately, that’s not always the case with every provider.
“Some providers, particularly those using open-source object detection not designed specifically for retail, can become burdensome in terms of the hardware retailers have to install,” Navarro said. “In some cases, retailers almost have to install what amounts to a server room somewhere in the stock room. But others, like ours, require just a small edge device and use relatively little computing power, even though they’re just as—and usually more—capable than those using off-the-shelf technologies.”
Traditional loss prevention solutions like hard tags, labels, and EAS detection systems aren’t obsolete—far from it. Computer vision simply adds a powerful new dimension to savvy retailers’ loss prevention ecosystems. It works alongside, rather than in place of—to augment existing technologies and empower associates to focus more on valuable customer interactions instead of watching monitors or looking out for suspicious behavior. And best of all, it’s relatively easy to implement and can be extremely cost effective, meaning it’s not a burden on your associates or your bottom line.
There’s a reason computer vision is the future of loss prevention—but you don’t have to wait for tomorrow to start seeing value from it. Start a conversation today.