Magnit began testing smart scales in convenience stores

2024: Testing smart scales in convenience stores

Magnet"" began to test "smart" scales in stores "at home." The technology that works on the basis neural network allows you to speed up the process of weighing goods and serving customers, as well as reduce the burden on personnel. The company already uses this solution in large format stores - Magnit Extra superstores and Magnit Family supermarkets. The company announced this on July 17, 2024.

"Smart" scales determine the product due to the built-in camera, which photographs the products placed on the weight platform. The photo is then sent to a recognition server, where the neural network analyzes the image and displays the most likely product options to the buyer. You can weigh fresh and frozen fruits and vegetables, sweets, nuts, snacks, frozen semi-finished products, seafood and other products sold for weight on smart scales.

The recognition accuracy is 98%, while the system is able to "recognize" products in transparent packages, as well as self-learn, remembering the choice of the buyer when weighing goods.

As part of the pilot project, "smart" scales were installed in the test group of Magnit stores "at home" in Krasnodar, Sochi, Anapa, Gelendzhik, the village of Golubitskaya (Krasnodar Territory), Volgodonsk, Tambov, Orel and Tula. The system recognizes the entire range of weight products, which in convenience stores averages 252 SKU. The test period will last until the end of September, then Magnit will decide on the possibility of scaling the project. When reaching the targets at the pilot stage, at the next stage, 2.8 thousand convenience stores can be equipped with "smart" scales - primarily with an expanded assortment of weight goods, such as vegetables and fruits, and in locations with high consumer traffic.

Magnit has begun testing smart scale technology in large format stores. In 2024, all supermarkets and superstors of the chain will be equipped with such scales.

{{quote 'author = said Eduard Iryshkov, Director of the Retail Technologies Department of the Magnit Chain. | Smart scales that recognize goods using computer vision have shown their effectiveness in hypermarkets. This technology allows you to increase the convenience of purchases and the quality of service: now the buyer does not need to remember the product code - just put it on the scales. In addition, the implementation of this solution in stores helps to increase the speed of service at cash desks, reduce the burden on personnel and reduce commodity losses. Now we want to test the technology of "smart" scales in stores "at home" - the most popular format of "Magnet." In recent years, this format has undergone significant changes: the importance of the service component has grown, the supply of fresh food, including vegetables and fruits, has expanded. Therefore, "smart" scales can become a popular solution in convenience stores, and we want to test this hypothesis in practice, }}

2023

A plan to scale product recognition technology to the entire store chain

On September 26, 2023, Magnit announced that by the end of 2023 it would scale the technology for recognizing goods on the shelf (Image recognition) to the entire retail chain. This solution allows you to quickly correct inaccuracies in the calculation, which helps to increase sales.

Control of the display of goods on the shelf is carried out using Russian technology, it is based on a neural network that recognizes images of goods by photograph and controls the compliance of the display of products on the shelves with the declared plan-scheme (planogram). The system analyzes the availability of the necessary goods and their balance in the store warehouse, the layout sequence, the location of the items on the first line and other indicators. If errors are detected for their prompt correction, the personnel receives detailed prompts to the mobile application.

Magnit began testing Image recognition technology in the fall of 2021 in 20 stores. A year later, it was decided to introduce this technology in a thousand outlets. Automated monitoring of calculations covered 20 thousand. SKU in 17 product groups. Shelf recognition accuracy exceeded 95%. This technology also contributed to an increase in trade turnover and the correctness of stocks in stores.

Now the company is starting to introduce technology for recognizing goods on the shelf in all Magnit stores of near-home, drogery formats, as well as in supermarkets and superstores. By the end of the year, this solution will work in more than 25 thousand outlets in all regions of the network's presence. At this stage, automated monitoring will cover 50 thousand product names. In addition, the technology will have additional functionality (for example, monitoring the correctness of price tags and monitoring advertising materials).

Magnit also began testing shelf-based product recognition technology at the Dixy retail chain.

{{quote 'author = said Ruslan Ismailov, Deputy General Director, Director of Magnit retail chain. | We scale the Image recognition technology to the entire Magnit chain of stores, which will allow us to more effectively control the physical availability of goods and compliance with the planogram, which means to improve the representation of goods on the shelf and increase sales. This project is one of the elements of the strategy for increasing the automation of routine processes in our stores, which allows not only to increase staff productivity, but also to obtain a large array of analytical information on commodity movement, an optimal planogram for a specific category in a particular store,}}

Plan to scale the technology of "smart" scales for 200 stores

On April 19, 2023, Magnit announced that it was scaling the technology of smart scales. Smart scales automatically recognize the product that the buyer weighs, which reduces the time spent on this process and improves the quality of service.

"Smart" scales determine the product due to the built-in camera, which photographs the products placed on the weight platform. Then the photo is sent to, server recognitions where neuronet it analyzes the image and displays to the buyer the most likely options for goods. By the end of 2023, this technology will be equipped with more than 1.5 thousand self-service scales in 200 large-format stores - supermarkets and superstores.

The system recognizes the entire range of weight products. This is more than 800 SKU in categories such as fresh and frozen fruits and vegetables, sweets, semi-finished products, seafood, snacks, etc. The technology is used for goods in transparent packaging (polyethylene bags, nets or avos). The recognition accuracy is 98%, while the system is self-learning, remembering the choice of the buyer when weighing the goods.

Magnit tested smart scale technology in 37 large format stores during 2021-2022. The pilot project showed a 0.3% increase in profitability by categories of weight goods due to a decrease in turnover. In 2023, the company plans to conduct a pilot project in convenience stores, equipping the scales used by cashiers with this technology.

Шаблон:Quote 'author = said Eduard Iryshkov, director of the retail technology department of the Magnit chain.

Magnit uses image recognition technology using a neural network to control the display of goods on the shelf: the neural network recognizes images of goods by photo and controls the compliance of the display of products on racks with the declared plan-scheme (planogram). The system analyzes the availability of the necessary goods and their balance in the store warehouse, the layout sequence, the location of the items on the first line and other indicators.

2021

Start scaling queue video analytics

On July 12, 2021, Magnit announced the beginning of scaling the video analytics of queues. The system allows you to quickly identify queues at the checkout and notify store employees about the need to open an additional one. Already in July 2021, the technology is installed in 60 Magnit Family supermarkets and Magnit Extra superstores, and by the end of September it will appear in 21 more stores.

The decision to scale was made in accordance with the Company's plans to minimize the occurrence of queues and make the purchase process even more comfortable for visitors. In this regard, the geography includes, first of all, stores with a large trading area and the greatest concentration of traffic.

Magnit installs video analytics in the cash register area, which includes both regular cash registers with cashiers and self-service cash desks. Cameras record a violation if there are invariably five or more buyers in the queue. The system sends an alert online to a chat in a work messenger specially created for employees. Upon receipt of the notification, the operating personnel must respond to the signal - open an additional cash desk, and the system will independently notify buyers about the possibility of going to a free cash desk.

Integration services are also supported by regional teams responsible for efficiency and client service. The system duplicates alerts to them if the signal is not worked out in the time allotted for redistributing the flow of buyers.

It is planned to equip the Magnit Family supermarkets and Magnit Extra superstores in, To Moscow,, Krasnodar,, Astrakhan,, Volgograd,, Bryansk,, and Izhevsk Kazan Rostov-on-Don Perm To Samara other Surgut To Ufa cities with the technology of video analytics of queues. During the testing of the technology, the Company has already managed to halve the number of queues, as well as increase turnover due to improved service.

Шаблон:Quote 'author = said Ruslan Ismailov, Deputy General Director, Director for Retail Network Management "Magnit."

Testing smart scales

On July 6, 2021, the Magnit retail chain announced the start of testing self-service smart scales, which automatically recognize the product by appearance. The pilot project was launched in the Krasnodar Territory - in six Magnit Family supermarkets - in Krasnodar, Yeisk and Tuapse.

In "Magnet" appeared scales that photograph and recognize products in packages

The technology saves customers time on independent weighing of goods on the trading floor. Smart scales recognize the product using a built-in camera that photographs products placed on the weight platform. After that, the system sends the photo to the recognition server, where the trained neural network analyzes the image and displays the most likely product options to the buyer.

As of July 2021, the recognition accuracy is about 95%, due to further training during the pilot, the company plans to bring it to 99%. Natural further training does not require additional actions from the buyer or store staff - smart scales remember the choice of the buyer when weighing the goods.

The system is able to automatically recognize goods in various transparent packaging. In particular, in a plastic bag, in a colored transparent bag, in a mesh or a bag. During the pilot, the system will develop and adapt to the recognition of goods in new types of packaging, for example, eco-bags, etc.

The system recognizes the weight product in the following categories - fresh fruits and vegetables, sweets, frozen vegetables and fruits, semi-finished products, seafood, snacks. Total - from 200 to 250 positions, depending on the store. "Smart" scales are adapted to the features of the weight range - the system can work with products indistinguishable from each other by external features, or, for example, recognizes the same unique product, the appearance of which may differ in different stores or regions.

As Pavel Orlov, Director for Innovative Development and Partnership of the Magnit Network, noted, the balance interface is simultaneously suitable for users who are used to and want to continue to use the old way - for this we left the dialing of the product number or searching manually, and for those who have already managed to evaluate the new automated way.

The pilot project uses technical solutions from Russian developers, together with whom we sought to create a qualitatively new service. To do this, we combined many years of experience of scale manufacturers, developers of video analytics systems, retail and the needs of customers. For example, the interface of self-service scales was jointly redesigned, which has become more understandable and easy to perceive, he said.

The company added that smart scales can improve purchasing experience, at least double the speed and quality of service by automatically recognizing goods. The buyer does not need to remember the product code and search for the desired position in the large list. In the near future, the retailer intends to start testing "smart" scales for cashiers in "at home" formats, which independently weigh goods for buyers, the Magnit network said in a statement.