M.Video-Eldorado Group expands AI in-store video analytics after pilot

Russian consumer electronics retailer, M.Video-Eldorado Group, has rolled out an artificial intelligence powered in-store video analytics solution.

This was developed in-house by the company’s data office.

Data from in-store cameras is processed by a neural network based on YOLO, a solution that can detect and identify multiple objects in an image. Cloud solutions and infrastructure based on the Raspberry solution are used to process the video stream.

The technology analyses the real-time data flow and can distinguish employees from visitors to "superimpose” data on the location of people on the sales floor.

The system currently focuses on three main scenarios: "lonely shopper", "checkout lines" and "store heat map".

It identifies when a guest is alone in a store and automatically notifies sales consultants using a chatbot. Lines at checkout counters are analysed using neural networks and personnel are informed when more checkout staff are needed.

The “store heat map” analyses the density distribution of store visitors by zones to study customer behaviour, evaluate product arrangement and choose the best spots for placing advertising materials.

Following a pilot launched in March, the number of “lonely shoppers” in participating stores decreased by 75%, and the customer acquisition ratio has tripled compared to similar stores that were not using video analytics.

The pilot was launched in several M.Video and Eldorado stores, and plans are afoot to expand it up to 50 stores by the end of the year.

M.Video-Eldorado’s Head of Video Analytics Maxim Shakura, comments: “The video analytics system that we developed in-house is far more cost-effective than any third-party solutions and is tailored to our needs.”

“We have been impressed with the results since it was launched a few months ago. Data analysis is enabling us to learn more about the customer and offer a more convenient and personalised shopping experience."