VietPOS.AI launches Shelf Vision AI module for large-scale retail chains

At the end of April 2026, VietPOS.AI officially announced the Shelf Vision AI module — an AI camera solution that recognizes products on supermarket shelves, counts quantities, and sends real-time out-of-stock alerts. This is one of the most complete modules since VietPOS.AI launched its enterprise AI division in 2024.

The Classic Problem of Chain Retail

For retail chains with 20 or more stores, controlling shelf stock levels has been a decades-long challenge. Traditional methods rely on store employees manually checking 2–3 times per shift — which is labor-intensive and fails to detect out-of-stock situations for fast-moving items in time. The result: lost revenue when customers can’t find products, promotional items being obscured, or shelf price tags not matching the POS system.

Shelf Vision AI solves this by installing IP cameras on shelves or ceilings, connected to a local AI processing server at the store. The system recognizes products by SKU through visual recognition, counts the remaining quantity on each facing, and compares it to the standard planogram.

Three Main Automatic Alerts

During operation, the system sends three types of alerts to the shift manager’s tablet and the chain-level dashboard:

1. Out-of-stock alert — when the product quantity falls below a configured threshold, including the exact shelf location for quick restocking by staff
2. Misplaced product alert — when a product is placed in the wrong position compared to the planogram, affecting the customer’s search experience
3. Price tag mismatch — when the shelf price tag doesn’t match the POS system price, detected via OCR reading of price labels

These alerts are sent with a latency of under 2 minutes from the moment the camera detects the issue, fast enough for staff to handle during their shift without needing periodic manual checks.

AI On-Edge Infrastructure — Enterprise Data Security

Unlike cloud-based Vision AI solutions, VietPOS.AI’s Shelf Vision AI runs entirely on local infrastructure at the store. Cameras connect to an edge server — typically an embedded GPU mini-PC — which processes images on-site and only sends metadata (quantities, alerts) to the central data center. Raw images never leave the store.

This deployment method meets enterprise data security requirements — especially for FDI clients or retail chains with strict data policies — while also reducing internet bandwidth costs compared to streaming video to the cloud.

Commercialization Roadmap

The module is currently being piloted at two major retail chains in Ho Chi Minh City since March 2026, across 80 stores. Pilot results show that store employees have reduced manual inventory time by 40%, while the rate of detecting out-of-stock situations within 15 minutes has significantly increased compared to the old process.

After the pilot, VietPOS.AI plans to commercialize in Q3/2026 with a complete deployment package: camera + edge server + software license + operational training. Interested customers can register for a demo at the website vietpos.ai.