We all know that security video cameras are becoming smarter. The IP cameras at the edge of today’s video surveillance systems contain computer chips that can potentially change how cameras are used. However, despite the changing technology and greater intelligence at the edge, today’s systems mostly use video cameras for one thing – to provide video. In some cases, the cameras provide hours and hours of video that no one will ever watch. 

Re-Examining The Role Of Video Cameras

Prism Skylabs is helping to drive a re-evaluation of the role of video cameras in the market. Founded in 2011, the San Francisco cloud service company thinks of IP cameras as sensors that are capable of providing a range of data that can be managed and processed in the cloud to provide more useful information to end-user customers. Prism’s current implementations of the “software as a service” approach focuses on retail merchandizing and marketing applications, but Prism Co-Founder and Senior Vice President Bob Cutting sees many other opportunities too. 

The cloud infrastructure provides a “reliable and continuously connected way to monitor and get data from cameras that is extremely robust and reliable,” Cutting says. Information from cameras “trickles up” to the cloud where data is “pre-extracted and available,” helping retailers optimize their store designs and marketing. Prism provides a blend of complementary visual and analytics data. 

"We looked at the camera and
reimagined what it can do as a real
sensor – a sensor with intelligence
that is cloud-ready, cloud-enabled
and easy to install. How we think of
video has to change"

Retail Applications Of Video Analytics

In the retail world, the approach enables marketers and merchandisers to constantly observe and monitor retail displays and customer activity from around the world in real time. Store owners can tell, for example, if their stores opened on time. Did a new product launch effectively? 

“There are hundreds of questions that retailers want to answer, and they don’t want to watch streaming video from the store,” Cutting says. “They just want answers to questions.” He says the system provides “an easy way to navigate and access data.” 

If you think of a camera as an intelligent sensor, the data provided by that sensor can take many different forms (and only one of them is “streaming video,” which may not be the most useful for a specific end user need). Cutting says the data is “privacy enabled,” and does not contain personal information.

Integrating With CCTV Manufacturers

Prism has integrated its cloud system with cameras from Digital Watchdog, and announced integration with Axis cameras at the recent ASIS International show in Anaheim. The company is also in the process of integrating with several other large camera manufacturers in the video surveillance market. Employing intelligent cameras at the edge, the company “saw overnight a shift from server-based solution to an edge-based camera solution.” 

“It’s the right form factor,” says Cutting. “We looked at the camera and reimagined what it can do as a real sensor – a sensor with intelligence that is cloud-ready, cloud-enabled and easy to install. How we think of video has to change.”

Role Of Video Analytics In Store Security

Security cameras are capable of providing up to a dozen additional outputs, combining data with visual elements, says Cutting. For example, intelligent cameras can count people, and can track movement of customers in a store based on defined rules. An end user can know how many people go down a certain aisle, how long they dwell in front of a display, how many people visit a certain area in a given time. 

Cameras can also provide “visual summaries” of activity in a store, showing graphically who went where over a certain period of time, providing retail traffic maps, heat maps, and other visual outputs to guide store owners and managers.

Cameras can also provide “visual
summaries” of activity in a store,
showing graphically who went
where over a certain period of time,
providing retail traffic maps, heat
maps, and other visual outputs to
guide store owners and managers

Finally, cameras can provide a variety of visual data (in addition to streaming video). These include video snapshots (high-resolution images taken periodically and delivered in high resolution to the cloud). Visual outputs might also include “background models,” which are images of retail shelves presented without the customers moving in front of them to provide a detailed view of products and how they are arranged on the shelf.

There are also other types of visual outputs, such as time-lapse video, and thumbnail images taken one frame per second. In effect, the visual output is matched specifically to what the end user wants to see – and one camera can be used for multiple outputs to meet the needs of various stakeholders. (Cameras can also provide outputs focused on the needs of loss prevention and security departments.)

Examples And Applications Of Retail Analytics

Lolli and Pops, a 26-store candy chain, is using the system to change the candy store experience. Using the Prism Skylabs system, the company tests multiple combinations of merchandizing displays, and measures the effectiveness (and maximizes the benefit) of each. The company employs A/B testing – one display in one store and a different display in a second store – to measure which approach works best, in effect fine-tuning the retail experience for customers. 

Another Prism customer is a large retailer deploying the system throughout Europe, leveraging the system’s ability to count, provide visual insight and understanding, and real-time visibility into the effectiveness of merchandizing displays (using a 25-point checklist to ensure compliance). 

Other potential end-markets include retail banking, hospitality and even casinos – “anyone who wants a better understanding of their space,” says Cutting. 

He says there is a growing opportunity for physical security integrators in the area of retail analytics, and use of cameras as sensors conforms to emerging industry trends such as “Big Data” and the “Internet of Things” (IoT). He asks: “How can we break down video into core components that are IoT-friendly and that a wider audience can use?”