Computer Vision in Retail: Lenta Reads Price Tags From a Moving Robot

In 2026, the retail chain Lenta (via its Lenta Tech team) turned the task of recognizing price tags from a moving robot's camera into a Computer Vision ML hackathon. The aim is to automate shelf monitoring and violation detection directly on the sales floor.

Why is this harder than a static camera?

Lenta mastered recognition from static cameras long ago and already uses it in its stores. A camera on a moving robot is a fundamentally different problem: frames blur, the angle and distance to the shelf keep changing, and the lighting in the store is uneven. As a result, reading price tags reliably on the move turned out to be noticeably harder to implement than the familiar recognition from a fixed point. This is the hard part the team had to solve before the scenario could work in a real store.

High-quality price-tag recognition from a moving robot's camera sounds

much more interesting and much harder, especially in implementation.

> — from the Lenta Tech corporate blog on Habr

What the task involves technically

Recognizing a price tag from a moving platform chains together several Computer Vision subtasks. First, the model detects the price tag itself in the frame, then reads its text: the product name, the price and the barcode. It also needs to handle motion blur and glare, and it must link the recognized tag to a specific spot on the shelf to work out which product it belongs to. On a static camera, a fixed viewing angle simplifies some of these steps. On a robot, every frame arrives from a new angle, so the model has to be more robust.

Which processes the scanning robot covers

The camera-equipped robot covers several in-store processes at once. Employees spend less time on manual rounds and shelf monitoring, while the robot flags violations as it moves, such as missing or incorrect price tags, and passes that information to the team. According to Lenta Tech, this removes routine work from staff and speeds up the response to problems with shelf display and pricing.

Key points from the publication:

• The task is recognizing price tags from the camera of a moving robot on the sales floor
• The foundation is Lenta's experience with static-camera recognition, which it has long used in its stores
• The goal is to cut the time employees spend on shelf monitoring and to flag violations automatically
• The format is a Computer Vision ML hackathon using data captured by the robot

What this means

Computer Vision in retail is shifting from fixed cameras to mobile robots: recognition on the move turns shelf rounds from manual routine into an automated process. Lenta's case shows that even a well-worked CV task has to be reinvented once the camera starts moving, and that businesses are willing to put such problems to hackathons to assemble a working solution faster.