Towards Safer Communities: Real-Time Violence Detection and Instant Alerts using mobilenetv2

Violent actions represent a serious danger to public safety and the overall well-being of society. Traditional surveillance systems operate primarily reactively, necessitating the development of proactive solutions to detect and mitigate such incidents. It is imperative for monitoring systems to identify violent behavior in real time and promptly notify the authorities. Deep learning and computer vision have emerged as pivotal areas of research for advancing automated violence detection systems.This paper proposes a Real-Time Violence Detection and Alert System that employs the MobileNetV2 architecture to ensure effective and accurate classification of frames. The system integrates Cloud Firestore for secure and scalable data management, providing a database that authorities can utilize for analyzing and investigating future violence cases. Additionally, it employs a Telegram Bot for immediate incident alerts. With 96% accuracy, the system demonstrates reliability and adaptability across various environments, offering a robust and scalable solution for public safety while contributing to proactive crime prevention.