RetailAI® Protect provides state-of-the-art AI for loss prevention at self-checkout and staffed lanes. Leveraging award-winning product recognition technologies, the system accurately identifies and stops common scan errors as they happen – including mis-scans and ticket-switching – while helping to protect customer privacy. The system offers industry-leading accuracy while being massively scalable for effectively unlimited SKUs and stores.
SCO terminals are rapidly replacing staffed lanes – but shrinkage is significantly higher at SCOs than at staffed lanes.
Shrinkage cost retailers $40 billion in 2018, and and losses are trending higher each year.
Two of the most common forms of scan fraud are mis-scanning and ticket-switching.
A customer might substitute their product's barcode with a less expensive barcode (ticket-switch) or avoid scanning the barcode altogether (mis-scan).
A high-definition camera positioned over the SCO station processes video of the transaction using a server running RetailAI® Protect.
RetailAI® Protect identifies the issue in near real time, pauses the transaction, and flags it for the attention of the store staff.
Alerted by RetailAI® Protect, the store staff greets the customer and provides assistance with on-screen guidance of what item was not scanned.
As per experience in actual deployments, results suggest that RetailAI® Protect can reduce shrink caused by mis-scans and ticket-switching by 80%+. The business case can be easily proven during POC.
The model of RetailAI® Protect is capable of recognizing up to 1 million+ products, in real-time, during video analysis to detect and stop ticket-switching and mis-scans activity with high performance.
RetailAI® Protect self-learns through Malong's novel weakly supervised learning. New SKUs are automatically learned during regular SCO usage without a separate training phase, even as packaging changes over time.