Passenger Flow Monitoring of Elevator Video Based on Computer Vision

Abstract

Passenger flow monitoring plays an important role in elevator intelligent monitoring and elevator operation status analysis. This study gives one of definitions for elevator passenger flow, and proposes a counting method of the passenger flow in elevator based on the fusion of computer vision techniques. This experiment calculates the passenger flow in real-time detection by dividing the task into three parts, which is detecting the door status using the background subtraction method, detecting the current floor with Support Vector Machine (SVM) algorithm and counting passengers by YOLOv3 (You Only Look Once) model. The results show that the proposed method can accurately measure passenger flow even when passengers are heavily obscured.