Metro minimises self-scanning costs and risks with Nomitri
Metro massively reduces implementation costs and risks of self-scanning by using visual AI. Software from start-up Nomitri runs directly on customers’ smartphone. While shoppers use the Scan&Go function of their mobile phones, the application evaluates camera images from the device. In this way, the retailer can match registered goods with the contents of the shopping trolley. In addition, Metro can send its customers relevant cross-selling and up-selling offers in real time based on their individual shopping behaviour.
Jörg Decker from Metro and Nomitri CTO Max Fiedler presented the innovative solution at the EHI Technology Days, which took place this week on 9 and 10 November at the World Conference Center Bonn.
Customer’s smartphone recognises whether products have been scanned
A holder is attached to the shopping trolley that can hold any smartphone. It is aligned in such a way that the camera of the mobile device can see at any time the products the customer puts in or takes out of the trolley. As with other self-scanning solutions, the customer has to scan the barcode of the products. However, the camera of his smartphone verifies in parallel that all goods placed in the trolley have actually been registered. In this way, the app serves as an intelligent article surveillance system. Even if it is not possible to ensure one hundred percent theft protection, the smartphone camera can match the contents of the shopping trolley with the recorded purchases to a certain extent.
Recording the goods while shopping enables Metro to create suitable cross- and up-selling offers in real time. Jörg Decker is convinced that this form of customer approach is more efficient than the evaluation of historical sales data. Despite the complex functionality, the application does not put excessive strain on the battery of the customer’s device. That is why Metro has refrained from installing a charger in the holder
Retailers do not have to invest in complex hardware
Nomitri’s application is based on neural networks. Using Deep Learning AI, the software analyses video data. Usually, this requires high computing power, which is often provided by cloud servers. Max Fiedler explained that the Nomitri software was developed to run in real time on any device, such as smartphones or tablets, without losing much of its accuracy. This allows the start-up to offer retailers a solution for autonomous self-checkout without having to invest in costly camera installations and the development of a cloud infrastructure. The application also guarantees the protection of personal data. These never leave the customer’s device.
Jörg Decker explained that three different variants of the Intelligent Shopping Assistant can be realised with the application. The first works according to the ‘Bring Your Own Device’ principle on the customer’s smartphone. Furthermore, the software can run on a tablet attached to the shopping trolley. Metro is already piloting both applications. In addition, the company is planning to test a fully equipped intelligent shopping trolley. However, this will not be ready for the market until next year.
Decker emphasised scalability of the solution. It is easy to install and to roll out. According to Fiedler, the solution can be set up at a new customer within three days. The implementation in further outlets then only needed one day each. Following a successful pilot, first Metro stores are currently being equipped with the Nomitri solution. Edeka retailer Lüning in Rietberg is also testing the application.