How a Russian ice cream company generates hot profits with artificial intelligence

This summer, a major ice cream manufacturer in Russia, Chistaya Liniya (Pure Line), integrated a new solution for shelf image recognition. This AI-powered technology, developed by the Moscow-based startup Inspector Cloud, allows Chistaya Liniya to audit coolers, analyze product and staff data, and ultimately drive sales.

For image recognition, ice creams are more of a challenge than a treat: the glass displays are often fogged or frozen, making it difficult to easily identify bar codes and collect information. 

Inspector Cloud, a developer of image recognition services for the retail industry, has found a solution to address this difficulty. This solution is now used in thousands of stores by Chistaya Liniya, one of Russia’s top five ice cream producers, according to the British firm Nielsen Holdings.

In the last 90 days, Chistaya Liniya has processed between around 450,000 shelf images. “Our image recognition accuracy has reached 98%,” said Alexander Berenov, CEO and founder of Inspector Cloud. The startup’s solution is now integrated with the ice cream producer’s salesforce management system.

The project focusing on monitoring ice cream distribution started in the early summer of 2018. Among other tasks, Inspector Cloud was required to recognize the SKU (stock keeping unit) or the machine-readable barcode through a foggy glass.

“Our solution had to identify Chistaya Liniya ice cream bars placed horizontally in a cooler with ice cream produced by other companies. The products could overlap with each other, making it hard to capture the image,” Berenov added. 

Sales representatives take pictures of the cooler after product placement to allow the system to accurately evaluate their work via image recognition. Before Chistaya Liniya integrated this solution, operators were able to manually analyze only around 5% of store photos.

According to the head of distribution, Oleg Luknitsky, Chistaya Liniya has obtained data on 120 employees who deliver ice cream to stores as a result of the project. “We can now fairly analyze their performance and calculate rewards for our field team,” Luknitsky said. 

“We reduced our expenses drastically, because we don’t have to use manual labor any more. We have also increased the volume of processed data from 5% to 100%,” Luknitsky added.

The Inspector Cloud shelf image recognition solution is used by brands such as PepsiCo, Philip Morris International, Henkel, Unilever and Schwarzkopf.