Machine Learning System Enters Shopping Centers

A machine learning system can help read customer moods — and catch shoplifters.

A machine learning system can monitor customers in stores
– The system can detect gender, approximate age, and customers' emotions
– The system provides data for optimization

The Czech company Blue Dynamic has unveiled its eMotion product, which counts visitors at a business location and recognizes their age, gender, and mood. It's a solution built on cloud services and machine learning from Microsoft. The system continuously calibrates and improves its output to predict customer behavior and composition as accurately as possible.

The machine learning system delivers detailed marketing data on visitor demographics and evaluates visitors' emotions

Whether it's a shopping center and its stores, a bank, a hotel, a trade fair, or even a stadium, the operator gets detailed marketing data on the makeup of their visitors. The AI evaluates the emotions that the services provided triggered in customers, at a given time and place.

It can even recommend the right clothing size for something you liked, or tell you whether you were satisfied at checkout. The retailer can then adjust how their staff behave, or their business processes, in response to customers' typical emotional reactions.

At the same time, merchants get insight into how customers typically move through the store and learn which spots actually get the most traffic. To evaluate changes, they also get heatmaps corresponding to different periods or to changes made along the way.

The data can also be used to track conversion in detail — that is, to determine how the number of people and the time spent near a given product corresponds to its actual sales. And, of course, it's possible to evaluate how a product's presentation performs based on the emotions it evokes in people. That makes it possible to keep a constant read on customers' current preferences.

The machine learning system also makes it easier to spot suspicious or unusual behavior, and so, for example, to flag a potential shoplifter. It also makes it easier for centers to keep track of unwanted individuals — or, conversely, VIPs. It continuously updates their profiles and shares them across stores, as well as alerting staff when these flagged individuals are on the premises.

The system can identify suspicious or unusual behavior, which can help prevent theft

There's also potential for integration with IoT — in its basic form, for example, showing customized advertising based on who's watching. And it gets even more interesting when the system determines that someone very similar — quite possibly the same person — looked at a given product some time before. That's when things really get interesting for the retailer.