ICL Services Presents Successful Pilot Of The Aisee Video Analytics System At Auchan Retail Russia

Evgenia Galtsova explained that deploying over two thousand self-checkouts across the retail chain was a deliberate strategic choice by the company. This trend has proven its effectiveness, positively impacting checkout line throughput and, crucially, customer satisfaction.

Sergey Shcherbakov detailed the technical implementation of this solution. The concept of the AISee system, developed by ICL Services, is based on synchronous analysis of two real-time data streams. “It must simultaneously compare decoded data from video cameras about what is happening at a specific checkout with the corresponding transactional data from that checkout — about scans, cancellations, and payments. Based on this comparison, the algorithms infer inconsistencies,” he explained.

The system was required not only to accumulate the findings, but to use them promptly. “In case of violations, this information needs to be sent back to the checkout as a polite warning for the customer, as well as to staff or security via messenger for real-time response. The system also generates analytical reports to assess the overall effectiveness of the system,” added Sergey Shcherbakov.

As a result, a modular platform was created, comprising ML services for processing video streams, object classification and tracking, a service for interacting with various self-checkout modules, an automated workstation for operators in the monitoring room / a security staff monitor at the store exit, as well as other services that analyze ongoing events.

The system was trained to detect three main loss scenarios. First, when not all items are scanned. This could be a missed item during scanning, an incorrect quantity entered, or a barcode being obscured for some reason. Second, the set of items does not match the receipt due to an incorrect item being scanned or a swapped barcode. And the third scenario is non-payment, when a customer completes scanning but does not make a successful payment before leaving.

During the pilot project at two hypermarkets, over 4 million shopping transactions were analyzed in real time. “The effectiveness of the security service in the self-checkout area has increased several times thanks to the AISee Video Analytics System. This allows us to combat losses caused by human forgetfulness or errors without increasing staff,” noted Sergey Shcherbakov.

Following the conference, TAdviser SummIT participants were able to discuss the system’s capabilities in more detail with the development team at the ICL Services branded booth, which also featured ICL.Retail, the integrated retail and café cashier solution, and Pro.GenAI, the GPT bot for user and client support.