Crowd Counting and Detection

Abstract

Crowd counting is a valuable tool in today's culture. It has a wide range of applications, including human identification, population censuses, and security. Hu man tracking is a one-of-a kind image processing approach with a bright future. As a consequence of developments in deep learning, artificial intelligence, and other technologies, the number of crowds has increased substantially in recent years. The goal of this research is to create software that can monitor objects (people) as well as handle lists and counts. Using YOLOv5 (You Only Look Once) technology with Pytorch, the targeting system recognises objects and uses Deep Sort for tracking and counting. Furthermore, unlike the popular Yolo object recognition engine. which detects all items at once, this system recognizes just the objects required by the user, which aids in system speed.