Suspicious Shopping Behavior Detection

AI-powered real-time detection of potential shoplifting cases

Trusted by

Shoplifting appears to be an unavoidable aspect of running a retail business, but the total costs to both the public and businesses are substantial

● According to the Centre for Retail Research, with an average shrinkage rate of 1.44%, the combined costs of crime in the U.S. and Europe amount to $60.673 billion ● In the United States, retail shrink is about 1.6% of revenue. The average shoplifting loss is over $450 per incident. ● Commonly stolen items such as cosmetics, baby formula, personal care items, razors and electrical goods are often stolen to order and resold by criminals. ● Shoplifters contributes to workplace violence with 88% of retailers stating that shoplifters were “somewhat more or much more aggressive and violent compared with one year ago”.

With Scylla's Suspicious Shopping Behavior Detection, you can receive instantaneous alerts to potential theft and prevent losses from organized retail crime and shoplifting at your premises.

How it works

Scylla's Suspicious Shopping Behavior Detection system is trained to identify when a shopper takes an item and attempts to conceal it in a bag or backpack.

Scylla analyzes real-time video surveillance from the store, excluding the cashier area (as customers are allowed to place items in bags after payment).

When Scylla detects a potential shoplifting event, it automatically sends alerts to all assigned endpoints, such as security guards' mobile devices, accompanied by a short video clip of the incident.

Scylla's Suspicious Shopping Behavior Detection system has been trained on a large dataset of shoplifting videos captured by surveillance cameras, ensuring exceptional accuracy and reliability.

It integrates seamlessly with existing store CCTV cameras and is optimized to process multiple video streams using a single GPU, providing real-time event tracking.

Download the whitepaper

What makes Scylla Suspicious Shopping Behavior stand out

Centralized solution capable of handling multiple camera streams simultaneously

Reliable accuracy of up to 97%

The AI-powered action detection and classification core enables the analysis of the behavior of almost anyone in view. Scylla AI utilizes a cascaded "detect->segment->assess" approach, making it scalable and applicable in challenging scenarios like observing distant individuals or crowded environments.

Book a demo

Shopping Mall Security Checklist

Download Scylla’s checklist for shopping malls on measures that can help the security team improve the safety of shoppers and retail staff and protect the mall’s value.

Learn more