Implementation of AI Solution for Armed Person Detection in an Existing Video Surveillance System for a Pawnshop Chain

Unfortunately, stores and stalls that process cash and have cash register equipment often get exposed to armed attacks. While financial losses can be endured, the loss of human lives can never be. This problem is especially true for pawnshops, which, in turn, may operate in disadvantaged neighborhoods. But this is their business and it has its own mission.

At the end of 2023, Softacom was contacted by a company that owns a large network of pawnshops (let’s call it Company). All the pawnshops were compliant with legal requirements. Video surveillance, an alarm system, and a panic button had been installed. But since the pawnshops were opened at different times and locations, there was no standardization in equipment and contractors.

The company decided to expand its in-house security department to address armed threats and improve employee safety. The department’s objective was to detect armed individuals in the pawnshop using the existing CCTV infrastructure, without requiring additional funding. It should be noted that the current systems could detect motion and sometimes identify individuals (although with frequent false positives), but could not reliably detect weapons, which was a critical limitation.

The Company turned to Softacom for our expertise in AI, particularly in computer vision and neural network training for physical security applications.