Russia's leading grocery retailer X5 Group has deployed a neural network to boost accuracy at self-service checkouts. Piloted across 80 Pyaterochka and Perekrestka stores in Moscow, the algorithm helps customers properly scan items and is preparing to expand chain-wide.

The innovative solution addresses multiple common errors, like unregistered products or mistaken item punches. The AI also prompts if goods are incorrectly placed on scales, recognizing quantities and categories of coffee, alcohol, lookalike produce, sweets and more. While currently distinguishing classes, the system will learn to identify specific product varieties over time.

When errors occur, the AI indicates the wrong scan and can summon staff if the customer needs assistance. The product image also displays onscreen. Employees can remotely monitor checkout zones via tablets, canceling or modifying transactions as needed.

In the Moscow pilot, the AI decreased checkout errors by 10%. Using installed cameras and computer vision, the algorithm analyzes video streams locally in real-time without cloud support. An extra camera broadens the scanning area coverage for enhanced recognition.

As X5's Director of Innovation Roman Timaev explains: "Improving technologies based on user feedback is key. To speed shopping and reduce incidents, we developed a unique algorithm that recognizes products faster than humans."

X5's patented self-checkout invention is already deployed in 6,000 stores, handling over 40% of traffic and saving customer and employee time. 15,000 units also feature biometric payment options.

By augmenting the retail experience via AI, X5 makes shopping smoother and more enjoyable. As Timaev notes, reducing friction drives loyalty and repeat visits.

The rollout highlights how AI-powered automation can boost convenience and productivity when thoughtfully implemented. With checkout friction a common retail pain point, X5's innovation sets a new standard for frictionless shopping through AI.