Advanced Security System using YOLOv3

Abstract

In this paper, authors considered the problem of Security system with high budget. The Security System proposed here is used to detect the objects or humans using YOLO. It can detect any kind of objects using deep learning based object detector in different positions and ranges from the camera. Here authors will apply a cascade of background subtraction and object detection that will describe the unauthorized zone and figure out the intruders by streaming each frame from the raspberry pi to laptop. The system is designed with the application of advanced computer vision algorithms that includes deep learning, object detection, and classification. Authors have proposed a system to provide advanced security using a resource constrained device known as Raspberry pi. The proposed system is capable of detecting objects and generates alerts also to provide advanced security system. Instead of limited resources on board, system is capable of doing complex computing via using Intel movidious NCS-2. Intel NCS-2 is a TPU that will perform all the complex computing on Raspberry pi like implementation of CNN and Deep learning modules.