Aldi Süd is piloting automatic recognition of fruit and vegetables at its self-checkout systems. Prof Dr Stephan Rüschen, Professor of Food Retailing and Head of Studies at the Baden-Württemberg Cooperative State University in Heilbronn, reports in a posting on social network LinkedIn how self-checkouts at an Aldi store in Cologne’s city centre automatically identify non-scannable items.
A camera attached to the side of the touchscreen uses computer vision and artificial intelligence to identify types of fruit or vegetables presented. If several items are eligible, the screen displays a short list for selection. Professor Rüschen confirms in his posting that the system worked well for all chosen products. He also described the process as better and faster than with other self-scanning checkouts.
The software comes from manufacturer Diebold Nixdorf. Its solution dubbed Vynamic Smart Vision / Fresh Produce Recognition makes it easier for customers to register fresh fruit or vegetables without a barcode. Aldi Süd’s POS software provider Gebit took care of the integration into the existing systems.
Items recognised four times faster
Using self-service systems, consumers often take a long time to find the desired item. The AI-based application is designed to solve this problem. Customers can register selected fresh produce in just a few steps. They place the goods on the scanner scale. The software analyses the camera image and automatically identifies the product. The screen then displays a short selection list or even the exact item. The customer selects the relevant item or enters the quantity for products that are charged by the piece.
According to a customer survey conducted by Diebold Nixdorf, it takes an average of 10 to 15 seconds for a customer to select the right fresh product from a selection list. With Smart Vision, the time is reduced to three seconds per item. This means item recognition will be four times faster. The manufacturer claims that transactions per hour can be increased by up to 40 per cent this way. Previously, around three out of ten customers with fruit and vegetables in their shopping basket refrained from using self-checkouts. Thanks to the simpler process, utilisation rate of SCOs could increase. Interventions by supervisory staff due to fruit and vegetables could also be reduced by up to 45 per cent.
Self-learning algorithm updates automatically
The AI model runs directly on the self-service system and requires no additional hardware installation. It is connected to the provider’s AI platform in the cloud via internet. This enables a self-learning mode and continuous improvement of the model. The algorithm updates automatically and has already been trained on hundreds of fruit and vegetable items. Store staff can add new items at one of the self-service systems within two hours.
AI also helps to prevent fraud. It recognises and analyses objects in the camera’s field of vision, including those in the vicinity of the scanner, for example customers’ hands. By automating item selection, accidental errors or malicious customer behaviour can be reduced. However, the system is not immune to customers choosing the cheaper variety from the selection list if, for example, the organic product has a higher price.
According to information of The Retail Optimiser, Aldi Süd is currently only testing the solution in selected stores. It has not yet been decided if the solution will be rolled out. In addition to the store in Cologne, Aldi Süd is also testing the solution among other locations also in a store in Aachen.
