Researchers at the Georgia Institute of Technology conducted an evaluation of a mobile camera system steered by an algorithm designed to reduce street crime. The study’s findings were published on a scientific platform, reflecting rigorous testing and peer scrutiny.
The neural network behind the system was trained on years of crime data from Warner Robins, a city in Georgia with approximately 66,000 residents. The dataset encompassed a wide range of incidents, from petty disturbances to more serious offenses, offering a realistic view of crime patterns in a mid-sized American town.
The core premise is that every city contains zones where criminal activity concentrates, yet the Warner Robins analysis of police reports revealed a tendency for crime to unfold within urban corridors rather than remaining static. By mapping past incidents to probable future locations, researchers could pinpoint high-risk areas with greater precision.
Using artificial intelligence to forecast where crimes are most likely to occur, the team repositioned smart cameras to new vantage points. Over a three-month period, video surveillance correlated with a notable decline in street assaults and burglaries in the target area, contributing to safer streets in that pilot zone.
Researchers noted that such technology could support crime prevention efforts in other cities and counties that have limited resources. The neural network was designed to integrate with existing street surveillance systems, minimizing the need for costly overhauls while expanding situational awareness for law enforcement and public safety agencies.
Earlier efforts in other regions explored neural network applications for different purposes, such as recognizing familiar faces to locate missing pets, illustrating the broad potential of AI-enabled visual systems across public-interest scenarios.