Customers: Hippopotamus, Wagon Network (formerly Hippopotamus)
Contractors: Inference Technologies, Crystal Service Integration Product: Video analytics (projects)Second product: Artificial intelligence (AI, Artificial intelligence, AI) Project date: 2021/01 - 2021/04
|
2021: Smart Scale Technology Experiment
Kemerovo trade "Hippopotamus" On May 26, 2021, the network announced that it had conducted an experiment on the use of smart weights technology - due to the analysis of images and deep neural networks ones, the product placed on the scales is automatically recognized. The solution was developed by LLC specialists "Interference Technologies" and works in integration with the cash system Set Retail installed at the retailer. According to the results of the pilot project, the system learned to recognize goods that are part of 92% of all store weight purchases.
The Behemag assortment has an average of 197 weight products: fruits, vegetables, sweets, nuts, dried fruits and more. The task of the cashier is to find the product in the hierarchical menu on the keyboard or in memory drive the PLU of 3-8 digits.
The system was installed at 5 cash desks of one of the chain's stores. The device is mounted on an existing weight platform, consists of a camera and a computing module, which transmit data to a common neural network.
Over 1.5 months, 25,000 photos of weight goods were collected. As testing progressed, the neural network used in the computer vision system trained and improved recognition accuracy. As a result, in 93% of cases, the product was determined correctly, in 98% of cases it was included in the top 5 offered by SKU. The system works even if the item is in a package.
"With the support of CSI, we continue to implement and test technologies that make the store modern and attract new customers. One of them is Smart Scales, an innovative development that we were interested in assessing, "said Alexei Pavlov, IT director of the Begemag network. |