AI, IoT Put Surveillance Cameras to Work in Customer Service

CCTV is not just for catching bad guys anymore.

The job description for those ubiquitous CCTV cameras is expanding from surveillance to customer service and marketing thanks to artificial intelligence (AI), machine learning (ML) and Internet of Things (IoT) devices.

The Arcules Intelligent Video Cloud platform, unveiled at Google Cloud Next 18 this week at the Moscone Center in San Francisco, provides brick-and-mortar retailers and other consumer-oriented establishments with a new AI-based sales tool.

The platform aggregates and analyzes video surveillance and IoT sensor data to identify customer interactions with products to help in not only identifying sales trends but also in providing individual customer service, according to Arcules, a Canon group company based in Irvine, Calif.

In a post on the company's Web site, Andreas Pettersson, Arcules CEO, offered a real life example of how the AI-powered video platform can work in customer service.

"One French bookstore applied behavioral analytics to its video feeds to analyze shopper movements and facial expressions as they perused the brick-and-mortar shop," he writes. "The analytics software looked for reactions like surprise or hesitation, notifying the store's employees when cameras detected those emotions. This allowed clerks to intervene and help frustrated customers before they took their business elsewhere."

Even in the more traditional surveillance mode, AI-powered video cameras can provide added functionality, Pettersson writes: "Similarly, convenience stores like Amazon Go employ cameras with advanced object recognition software to confirm which items shoppers pick and ensure proper payment. Powered by deep learning technology, Amazon's store cameras use complex pattern recognition to detect when products are taken off the shelf. The cameras can even track when a customer put objects back. While the cameras double as a means to prevent shoplifting, their primary function is to improve the customer experience and support Amazon's mission of providing unmatched convenience for shoppers."

The advances in CCTV technology are due to two factors, according to Pettersson.

  1. Advanced object detection: With previous technology the cameras were only able to do limited object detection, such as distinguishing between a person walking and a car driving. "Deep learning algorithms now enable video monitoring systems to figure out specific details about what cameras are seeing. This includes more granular information such as if the person in the video is a woman or a man, and what color his or her clothes are," he writes.
  2. In-depth behavioral analysis: Previously, human operators had to view video recordings to try to figure out emotions or reactions from facial expressions caught on camera. "Analytics technology can now teach cameras how to read micro-expressions, helping marketers and behavioral analysts alike understand how their customers feel during their experience. Smart cameras, for example, can determine if a shopper is excited or confused when presented with different retail advertisements or product displays."

The breakthroughs producing the Arcules video platform, according to Pettersson, comes from technology moving beyond the rule-based algorithms that have provided video analysis in the past to now making use of AI and machine learning so "cameras get a 'brain.'"

Arcules' announcement of its platform said that it is "powered by technology from Milestone Systems, the global industry leader in video management software." Milestone XProtect is open video management software (VMS) with published APIs that allow developers to modify the platform for their own projects, according to the Milestone Web site. "The Milestone open platform enables you to add custom, best-in-class security solutions to your surveillance, such as access control, cameras and video analytics," the company explains.