Pedestrian Crowd Detection and Segmentation using Multi-Source Feature Descriptors

Saleh Basalamah,Sultan Daud

doi:10.14569/ijacsa.2020.0110187

Abstract

Crowd analysis is receiving much attention from research community due to its widespread importance in public safety and security. In order to automatically understand crowd dynamics, it is imperative to detect and segment crowd from the background. Crowd detection and segmentation serve as pre-processing step in most crowd analysis applications, for example, crowd tracking, behavior understanding and anomaly detection. Intuitively, the crowd regions can be extracted using background modeling or using motion cues. However, these model accumulate many false positives when the crowd is static. In this paper, we propose a novel framework that automatically detects and segments crowd by integrating appearance features from multiple sources. We evaluate our proposed framework using challenging images with varying crowd densities, camera viewpoints and pedestrian appearances. From qualitative analysis, we observe that the proposed framework work perform well by precisely segmenting crowd in complex scenes.

Highlights

With the growing population of the world and with the increased urbanization, scientific community focus on developing tools and techniques to ensure crowd safety
In order to ensure public safety in such high density situations, surveillance cameras are installed in multiple locations providing coverage of whole crowd scene
Despite the recent advancements in computer vision technology, research community still did achieve desired result for understanding crowd dynamics. This attributes to following reasons: (1) Most of existing methods are based on assumptions often violated in real world scenarios

Summary

Introduction

With the growing population of the world and with the increased urbanization, scientific community focus on developing tools and techniques to ensure crowd safety. In order to ensure public safety in such high density situations, surveillance cameras are installed in multiple locations providing coverage of whole crowd scene. This is the job of security personnel to detect abnormal activities by watching over the TV. Despite the recent advancements in computer vision technology, research community still did achieve desired result for understanding crowd dynamics This attributes to following reasons: (1) Most of existing methods are based on assumptions often violated in real world scenarios. These tools compute important measurements that are of significant importance to crowd managers and security personnel These measurements include but not limited, crowd counting, density estimation, anomaly detection, crowd tracking. These tool sets computes important measurements that are useful in understanding crowd dynamics yet these tools could not detect detect crowds in the scene

Methods

Results

Conclusion