From the thesis: "Mass shootings in the United States have become tragically commonplace. Since the notorious Columbine High School shooting in 1999, America has experienced a growing number of mass shootings across a variety of locations, targeting a wide range of victims and employing individual shooter tactics and approaches. While the best countermeasure against active shooters remains dispatching police to the scene of the attack as quickly as possible while urging intended victims to run, hide, or fight, this loose guidance for both victims and responding police provides the shooter with the initial advantage. Modern technology, however, may provide new approaches to address the active shooter problem. Enhanced sensory equipment, such as motion-tracking cameras, paired with powerful artificial intelligence (AI) models, offers an opportunity for engineers and computer scientists to build a cohesive suite of automated countermeasures to identify and potentially stop an active shooter attack at the onset. This thesis proposes an AI-driven surveillance system using object recognition to detect weapons and threatening behaviors at the onset of an attack. Based on case study analysis and simulation testing, the system offers a privacy-conscious approach to early intervention and casualty reduction in active shooter scenarios."