A machine learning based weapon detection system for enhanced security surveillance

Abstract

The present invention is related to a machine learning based weapon detection system for enhanced security surveillance. The invention is a cost-effective, real-time weapon detection system that utilizes the Raspberry Pi 4 platform and advanced machine learning techniques to identify weapons in video streams. By integrating the YOLOv3 object detection model with a custom-trained convolutional neural network (CNN), the system ensures accurate classification of objects as weapons or non-weapons. It operates efficiently with low power consumption (5 watts) and processes video at 15 frames per second (FPS), making it suitable for deployment in a variety of security applications such as public spaces, educational institutions, critical infrastructure, and law enforcement. The system is scalable, energy-efficient, and provides real-time alerts, enhancing security surveillance while reducing costs compared to traditional solutions.