Embedded machine-learning technology stops thieves passing off valuable items as inexpensive ones at supermarket self-checkouts.
The introduction of self-scanning systems in supermarkets has more than doubled the rate of shoplifting, according to research by the UK’s University of Leicester. Its analysis of 12 million shopping trips across Europe and the US shows that theft and fraud levels jump on average by 122% at supermarkets fitted with self-scanning technologies, pushing related losses to a painful 3.9% of company revenues.
Understandably, retailers have been hungry for solutions that can reduce such an impact, which in most cases results from fraudsters swapping barcode labels from lower-cost items to expensive ones or passing off costly items such as self-weighed fruit or vegetables.
Now, one trial by a major European supermarket chain is using new AI-based technology from Fujitsu to deliver a solution with the potential to save retailers billions annually.
The system works by capturing images of an item as it is presented at a self-scanning device, identifying the type of product from its visual characteristics and then crosschecking those against the item as its barcode is scanned. If a product’s expected
appearance fails to match the barcode, the self-checkout requests a manual intervention to determine next steps.
This AI-based fraud prevention solution complements existing self-checkouts, as its cameras are fitted to existing scanners. As such, it does not involve the cost and complexity associated with other options for identifying mismatches such as graphical processing units (GPUs).
Says Bruno Sirletti, head of retail and hospitality at Fujitsu EMEIA: “The technology makes it very difficult to trick the checkout, and achieves this without inconveniencing honest shoppers.” As such, it is a major step towards the viability of frictionless,
Part of Fujitsu’s Zinrai portfolio of AI solutions, the system is one of the first commercial applications developed in the company’s AI Center of Excellence, which was established in 2017 just outside Paris.