Checkout AI in action

Selling non-barcoded products — fruits, vegetables, baked goods — through self-checkout has been a long-standing issue for retailers. On average, picking a single item from the visual picklist menu takes anywhere from 6 to 30+ seconds, depending on the product type. The friction frustrates customers and generates avoidable shrinkage from incorrect product selection.

To tackle this, Retellect teamed up with StrongPoint, a leading retail technology company and self-checkout vendor. All four self-checkout machines in the selected Norfa store were equipped with cameras and connected to the Retellect AI service via StrongPoint integration.

The system has been running silently since September 2020, learning from real shopper behaviour before the public announcement.

Results

After more than six months in a live store environment, the deployment delivered measurable improvements for both customers and operations.

~2s
Average picklist selection time per non-barcoded item.
6+ mo
Live in production before the official announcement.

Picklist time dropped from a wide 6–30s range down to roughly two seconds per item. The solution proved reliable and robust in a Norfa store handling a wide range of fresh products — exactly the environment where conventional picklists struggle most.

AI bounding-box identifying a tomato among other produce.
Object detection at the lane — the AI locates the produce in frame and labels it in well under a second.

Seeing the benefits delivered through the Retellect and StrongPoint partnership, Norfa is now expanding the deployment to additional stores.

This was the first production deployment of Retellect's checkout vision platform — and the validation point that everything since has been built on.