Morrisons Deploys Everseen Vision AI Across 200 Stores
Morrisons has selected Everseen to deploy its Evercheck computer-vision platform across 200 UK stores in an initial rollout, following a trial. Retail Systems reports that the system analyzes self-checkout camera feeds to identify products and potential missed scans, while on-screen prompts let shoppers correct exceptions. Morrisons' Store Support Office Director Francesca Bezoari said the partnership is intended to support faster checkout and customer service.
Morrisons has selected Everseen to deploy the vendor's Evercheck computer-vision platform across 200 UK stores in an initial rollout, following a trial of the technology. The deployment targets self-checkout lanes, where real-time vision models analyze activity around scanned and bagged goods.
According to Retail Tech Innovation Hub and Mass Market Retailers, Evercheck presents on-screen prompts that allow customers to correct missed items themselves. The publications report that the software integrates with existing store systems and is intended to reduce the need for store-team intervention during checkout exceptions.
"Our new partnership with Everseen enables us to provide the fast and seamless checkout experience our customers want, while simultaneously equipping our colleagues to maintain the exceptional service that Morrisons is known for," said Francesca Bezoari, Morrisons' Store Support Office Director, in coverage published by Retail Tech Innovation Hub and Mass Market Retailers.
Computer vision at the checkout
Retail Systems reports that Evercheck uses AI to recognize products through self-checkout cameras, including fresh produce, and can flag cases in which an item is bagged without being scanned. The publication also reports that the technology can identify product-selection mismatches, such as a shopper selecting a lower-priced produce item for a higher-priced one.
Everseen CEO Joe White said in Mass Market Retailers that the platform enables Morrisons to improve satisfaction scores while maintaining operational performance. Everseen's assertions on visual-recognition accuracy and operational results were not independently benchmarked in the retrieved reporting.
The 200-store rollout is a material retail-computer-vision deployment because it places model inference in a high-volume, customer-facing workflow rather than solely in back-office loss-prevention operations. In comparable deployments, practitioners typically have to balance detection quality against false-positive prompts, latency at the lane, edge-camera coverage, and escalation paths for ambiguous events. Customer-facing prompts can lower staff workload when accurate, but excessive or poorly timed interventions can create friction at precisely the point a self-service system is meant to accelerate.
Operational and governance considerations
Retail Systems frames the agreement as addressing both checkout speed and retail losses. For teams evaluating similar systems, the reported use cases illustrate why checkout vision deployments require more than image classification: the system must associate camera observations with point-of-sale events, product catalogs, weight or bagging signals where available, and a workflow for resolving exceptions.
Morrisons has not published technical details in the retrieved coverage on Evercheck's model architecture, inference hardware, performance thresholds, data retention, or the rate of human review. Those details would be important for assessing model reliability, privacy controls, and whether performance differs across produce categories, store layouts, lighting conditions, and shopper behavior.
The rollout also adds to the use of AI-assisted exception handling in physical retail. Industry deployments of this kind commonly concentrate value in reducing manual interventions and preventing avoidable losses, while requiring careful measurement of error rates and customer experience across diverse store environments.
Key Points
- 1Morrisons is deploying Everseen Evercheck across 200 stores, extending computer vision from a trial into a substantial self-checkout implementation.
- 2Reported capabilities link camera observations to checkout exceptions, including missed scans and produce-selection mismatches, rather than operating as standalone image recognition.
- 3Comparable retail vision deployments depend on low latency, low false-positive rates, robust point-of-sale integration, and clear human exception-handling workflows.
Scoring Rationale
This is a notable production deployment of computer vision in a high-volume retail workflow, with direct relevance to teams building physical-world AI systems. The reporting does not disclose technical performance metrics, model details, or privacy controls, limiting its broader research and infrastructure significance.
Sources
Public references used for this report.
Practice with real Payments data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Payments problems
