In a contributed article for Internet of Business, Katie McCann, vice president of product and engineering at Prism Skylabs, discusses how the combination of visual data collected by surveillance cameras with artificial intelligence (AI) could lead to better in-store experiences for shoppers.
Countless research studies warn of the damaging effects on physical and mental health associated with living too much of our lives online. The online trend is also affecting the health of organizations in industries like retail, healthcare, education and finance, which tend to use physical, ‘real world’ spaces to deliver at least some of their goods and services. Could the drift to online mean that public spaces like high streets and shopping centers soon become ‘ghost towns’?
This dystopian viewpoint overlooks humans’ very strong basic psychological and physiological urges to touch, feel and socialize. It also overlooks the fact that the IoT is making the internet physical, so that fridges and watches get connected and become smarter. The IoT is also enhancing familiar experiences, from shopping to driving to going to the theater, by combining the best aspects of the virtual and physical worlds.
Let’s consider how the IoT augments the Internet Protocol (IP) surveillance cameras we see in retail environments. Shoppers assume – and they are largely correct – that these cameras are installed for loss prevention. But, as my colleague Cliff Crosbie told Internet of Business almost a year ago, by adding software to in-store, IP cameras can become data-gathering IoT sensors.
They can count shoppers accurately, provide data on the paths they take through retail experiences, show where they linger. That data gives store employees the insights they need to improve the store environment for customers. At Prism Skylabs, we’ve even figured out a way to gather all this data while simultaneously stripping out people’s identities from video footage. This removes the ‘Big Brother’ aspect, while giving physical retailers all the advantages of data normally only available to their online counterparts, in the form of clickstream analysis.
But using cameras to sense people’s presence and location in retail spaces is just the beginning. Artificial intelligence (AI) technologies and communities of developers are now providing the algorithms to give cameras the power to do so much more. The famous neuroscientist Oliver Sacks wrote in one of his last books, Hallucinations: “One does not see with the eyes; one sees with the brain, which has dozens of different systems for analyzing the input from the eyes.” Much like this, I’m part of a team working to apply AI technology (the brain) to analyze the input from cameras (the eyes).
AI as ‘force multiplier’ for video IoT
New AI developments will help retail and other service industries tap their cameras’ data to make physical spaces better in every way – easier to navigate, safer, more profitable, and more fun to be in. That visual data is already being gathered, but because it requires so much focused human effort to analyze, it’s mainly been used to investigate exceptional events, like break-ins. AI can act as a true ‘force multiplier’, however, making digging through loads of video data easy and efficient. That means that, in a challenging sector like retail, video IoT’s use can stretch well beyond security and loss prevention to:
Enhance and protect the brand. Today, physical retail spaces are increasingly used as ‘brand showrooms’. These benefit from intelligent visual information that enables store managers to maintain the right standards when it comes to cleanliness, style, staffing and other factors. In other settings, such as a hospital, this takes on the added importance of protecting patients’ health.
Build context for decision-making. Retail store managers need a long history of data about things like visitor numbers at different times of day/week, in order to be able to make smart decisions about promotions, for example, and to negotiate favorable agreements with suppliers. AI is making it possible to collect more granular information like the gender, height and even preferred brands of store visitors, while still protecting their identities. Building context is really difficult in many industries, especially those with high levels of staff turnover and temporary/contract employees, such as retail, healthcare, construction and manufacturing.
Diagnosing problems. In retail, knowing where the ‘hotspots’ are in a store is important. Even more important is identifying the ‘coldspots’ that customers don’t visit. This enables merchandisers to relocate promotional displays and items that are performing poorly and also diagnose and fix problems with store locations. This is already possible today, but AI has the potential to support this further by immediately detecting and alerting staff to common problems like poor lighting.
These are only a handful of examples that illustrate the vast potential for applying AI to enhance video IoT. As I work with our own development team – and the wider community as we open up our APIs [application programming interfaces] – one of my biggest challenges is prioritizing among the many areas of innovation.
The most satisfying part, however, is being able to see the positive way video IoT is already starting to support radically new ways of using spaces. Creative retailers like our customer STORY – a New York-base store that completely reinvents itself with a new theme every four to eight weeks that covers everything from store design to merchandise – are applying video IoT and other digital technologies to reignite people’s natural enthusiasm for the physical shopping experience.
We are also working with early adopters to apply video IoT in other sectors like healthcare, security and construction. So despite the current difficulties retailers and other industries are facing, I not only see technical solutions under development but practical applications starting to make a real difference.