The goal of this Ready Solution is to prevent fraud as it occurs without affecting the customer experience. It is a self-learning computer vision solution that can protect a wide range of stock keeping units (SKUs) by detecting mis-scans and ticket switching in near real time. The main design criterion for choosing solution components is determining if they can co-exist with and augment existing POS scanners.

Two scenarios in which the Malong RetailAI Protect solution can prevent retail loss are ticket switching and mis-scans ( see Traditional methods for reducing inventory loss). An overhead fixed-dome camera captures video of an item that is either not scanned or has a suspect UPC barcode. The video is decoded in the GPU and the decoded video is used as input to the Malong RetailAI model. The model then predicts the item's UPC barcode. If Malong RetailAI Protect determines that:

• The item was never scanned, after a set interval of time, Malong RetailAI Protect notifies the SCO system to raise an alert
• The scanned UPC barcode does not match the correct UPC barcode, Malong Retail AI Protect immediately notifies the SCO system to raise an alert

The retail associate overseeing the SCO lanes can then take the necessary action.

This solution is easy to put into use at scale for thousands of retail stores. The technology for this solution can be used at both SCO and staffed lanes.