Enhancing Safety at Over Crowd Places Using Modified YOLOv8

 Authors:

1. Subramanyam¹, P. Prathap², K. Suneel Kumar³, M. Rajasekhar⁴, Y. Lokesh⁵, Mr. P.

Venkateswarlu⁶

^1,2,3,4,5B.Tech Students, Department of Electronics and Communication Engineering, PBR Visvodaya Institute of Technology & Science, Kavali, Andhra Pradesh, India

⁶Associate Professor, Department of Electronics and Communication Engineering, PBR Visvodaya Institute of Technology & Science, Kavali, Andhra Pradesh, India

Abstract - Managing crowd safety in overcrowded public spaces is a critical challenge due to high population density and dynamic crowd movement. This paper presents a real- time weapon and fire detection system using the YOLOv8 deep learning model deployed on a Raspberry Pi. A USB camera continuously captures live video to detect weapons, fire hazards, and overcrowding, while face detection provides visual evidence of individuals involved. Upon threat detection, the system automatically captures an image, retrieves GPS coordinates, and sends an email alert with location details to authorities. Local alerts are simultaneously triggered through a buzzer, speaker, and LCD display. The system combines edge computing, computer vision, and automated communication to enable faster emergency response and reduce dependency on manual surveillance in high-density environments.

Key Words: YOLOv8, Raspberry Pi, weapon detection, fire detection, crowd monitoring, GPS alert, real-time surveillance.