Back to search

University of West Attica

Real-time detection, evaluation and mapping of crowd panic emergencies based on geo-biometrical data and machine learning

Abstract

dc:description

In recent years, the demand for efficient real-time emergency response systems (ERS) has increased substantially, largely due to the urgent need to detect, map, and respond to crowd panic situations that pose significant risks to public safety. In this study, we propose an innovative, real-time, geo-referenced ERS that leverages biometric and spatiotemporal data collected from wearable devices and smartphones to provide actionable intelligence in emergency scenarios. Unlike traditional visual surveillance technologies and community engagement systems, the proposed ERS employs advanced machine learning techniques to process incoming data. In particular, a Gaussian Support Vector Machine (SVM) classifier is utilized to differentiate individuals experiencing panic from those exhibiting normal behavior , thereby introducing the Stress Profile Index (SPI). By enabling real-time data collection and analysis, the system seamlessly transmits critical information from the source of the event -i.e., the human body- to the appropriate emergency response teams. Furthermore, this research introduces the Domino Effect Index (DEI), a novel metric designed to gauge the severity of panic-induced crowd behavior . The DEI considers multiple factors, including the rate of panic propagation, the density of affected individuals, and the alignment of their movement with the surrounding road network. By incorporating the DEI, emergency responders can quickly evaluate the severity of an ongoing event and enact effective measures to prevent escalation, thereby enhancing public safety. Real-time spatial analysis, performed using labeled subjects’ geospatial information, is integral to this framework. Through such analysis, it becomes possible to detect latent patterns that may indicate the onset of crowd panic, thereby providing essential insights to first responders. This analytical approach further facilitates the visualization of regions where aberrant crowd behavior is emerging, supporting proactive mitigation efforts. The proposed ERS continuously generates dynamic areas to represent the real-time spatial development of crowd panic emergencies. These dynamic areas offer responders an immediate overview of the location and progression of the incident, enabling them to calibrate their interventions for maximum efficacy. Three types of dynamic areas are produced: one corresponding to the initial stressed locations of individuals, another reflecting their current stressed positions, and a third illustrating the geometric difference between the two—a measure that captures the evolving nature of the emergency. By integrating both machine learning methodologies and the DEI metric, the proposed ERS emerges as a robust, dependable, and efficient platform for the detection, mapping, and severity assessment of crowd panic emergencies. The ability to process real-time biometric and geospatial data empowers emergency responders to develop more effective strategies and optimize resource allocation, ultimately strengthening public safety. In conclusion, our real-time, geo-referenced ERS—enhanced by the inclusion of the DEI metric—constitutes a transformative approach to managing crowd panic emergencies. By enabling the rapid detection, spatial mapping, and severity assessment of such events, the system equips emergency responders with the information necessary for targeted interventions, thereby improving public safety and mitigating the risks associated with panic-induced crowd movements.

Degree

thesis:*
Grantor dc:publisher
University of West Attica
Year dc:date
2025

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Lazarou, Ilias
  • Λαζάρου, Ηλίας

Subjects

dc:subject × 19

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/59752

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Lazarou, Ilias; Λαζάρου, Ηλίας. Real-time detection, evaluation and mapping of crowd panic emergencies based on geo-biometrical data and machine learning. University of West Attica, 2025. http://hdl.handle.net/10442/hedi/59752