Quantifying Risk in Pedestrian Crowds Using Divergence Estimated from Flows of Head-Tracking Data


Haruto Nakayama (University of Tsukuba), Masaki Onishi (AIST)
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)Supplementary

Abstract

The rapid pace of urbanization and the increasing frequency of large-scale events have made crowd management a critical concern in urban planning, facility design, and event operations. Effective crowd management must address congestion while ensuring pedestrian safety and comfort. Achieving this requires a reliable assessment of potential risks within crowds. While crowd density is a common indicator of congestion and can identify hazardous high-density situations, it is inadequate for evaluating risk in low- to medium-density crowds, where pedestrian movement dynamics play a more significant role. To overcome this limitation, we propose the crowd risk score (CRS), a novel quantitative metric for assessing crowd-related risk grounded in the mathematical concept of divergence from vector calculus. The proposed approach was validated using datasets derived from real-world event footage with human-annotated risk labels. Experimental evaluations on these datasets demonstrated the effectiveness of the proposed method. By computing the CRS from pedestrian data automatically obtained through deep learning-based detection and tracking, the proposed method enables end-to-end analysis of crowd-related risk directly from video input. This automation supports crowd monitoring and analysis in practical crowd management scenarios.

Citation

@inproceedings{Nakayama_2025_BMVC,
author    = {Haruto Nakayama and Masaki Onishi},
title     = {Quantifying Risk in Pedestrian Crowds Using Divergence Estimated from Flows of Head-Tracking Data},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_1024/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection