AI-DRIVEN SURVEILLANCE GUN DETECTION

Abstract

Unpredictable and growing threats are contributing to the ongoing rise in security concerns in public areas.
Because they prioritize recording video for subsequent analysis over enabling instant response, traditional
surveillance systems are frequently inadequate. This project makes use of the cutting-edge YOLOv8 object
detection model in an AI-based surveillance system for identifying weapons. It uses thermal imaging and
RGB to provide dependable detection in a variety of lighting environments. The system processes highresolution
video locally using edge computing, which minimizes latency and eliminates dependency on
cloud infrastructure.The system also includes a user-friendly and secure web-based control panel that is
created using Bootstrap 5 and FastAPI. With this interface, security employees can upload video data,
monitor live analysis, and effectively examine identified risks. The system immediately notifies via SMTP
email when a weapon, such as a rifle or pistol, is discovered, allowing for rapid communication and quicker
response times. With this method, conventional surveillance moves from passive monitoring to an
intelligent system that can identify and react to threats in real time. Through ongoing surveillance, precise
identification, and quick notification production, the system enhances public safety when used in crucial
sites like public gatherings, transportation hubs, and other vulnerable areas