essay about crowd management

Place crowd safety, crowd science? Case studies and application

Journal of Place Management and Development

ISSN : 1753-8335

Article publication date: 16 April 2020

Issue publication date: 22 September 2020

This paper aims to explore the development and application of place crowd safety management tools for areas of public assembly and major events, from a practitioner perspective.

Design/methodology/approach

The crowd safety risk assessment model is known as design, information, management-ingress, circulation, egress (DIM-ICE) (Still, 2009) is implemented to optimise crowd safety and potentially throughput. Three contrasting case studies represent examples of some of the world’s largest and most challenging crowd safety projects.

The paper provides some insight into how the DIM-ICE model can be used to aid strategic planning at major events, assess potential crowd risks and to avoid potential crowd safety issues.

Practical implications

It provides further clarity to what effective place management practice is. Evidence-based on the case studies demonstrates that the application of the DIM-ICE model is useful for recognising potential place crowd safety issues and identifying areas for require improvement.

Originality/value

Crowd science is an emerging field of research, which is primarily motivated by place crowd safety issues in congested places; the application and reporting of an evidence-based model (i.e. DIM-ICE model) add to this. The paper addresses a research gap related to the implementation of analytic tools in characterising place crowd dynamics.

Place crowd safety
Crowd science development

Still, K. , Papalexi, M. , Fan, Y. and Bamford, D. (2020), "Place crowd safety, crowd science? Case studies and application", Journal of Place Management and Development , Vol. 13 No. 4, pp. 385-407. https://doi.org/10.1108/JPMD-10-2019-0090

Emerald Publishing Limited

Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial & non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode

1. Introduction

Managing crowded places is challenging. Over 100 years ago, in 1902, 25 people died and 517 were injured when the West Stand at Ibrox Park, Glasgow, UK collapsed during an international football match ( Still, 2019 ). In 2019, 16 people died and 101 were injured in a human crush in Antananarivo, Madagascar before a concert at the Mahamasina Municipal Stadium. The show was about to start and people, believing they could enter the stadium, began to push, however, the doors were still closed ( Still, 2019 ). Recently we have global climate strikes on 6 December 2019, with disruptive demonstrations in 2,300 cities across 153 countries ( Al Jazeera, 2019 ). Failures in both place design and place management are not unique and crowd “mis” behaviour is not always the primary cause of accidents and incidents. One common factor is the inappropriate utilisation of space. After major incidents when facts are analysed the crowd is rarely the cause. More commonly the design and management of the place is the problem.

The safety of humans in crowded environments has been recognised as a rapidly growing research area and has been of significant concern to many government agencies ( Helbing et al. , 2007 ). Increases in urban populations and mass events have raised interest among researchers and authorities in regard to the problems of pedestrian and crowd dynamics ( Haghani and Sarvi, 2018 ). To date, there has been limited empirical research on pedestrian and crowd behaviours, dynamics and motion ( Shahhoseini et al. , 2018 ). Identifying and understanding the mechanisms that may lead to crowd disasters and incidents are critical to ensuring safety in crowded environments ( Helbing et al. , 2007 ). In addition to this, place management aims to identify and understand elements such as the political, legal, economic, social and technological aspects of our environment, which ideally lead to ensuring it is “fit for purpose” ( Kalandides et al. , 2016 ; Parker, 2008 ).

This paper reports on the implementation of a crowd safety management tool for places of public assembly and major events and is a practitioner, not a conceptual paper. It provides insight into strategic planning for places regard major events, specifically how to potentially reduce crowd safety issues and as such makes a contribution to the place management and development literature ( Badiora and Odufuwa, 2019 ; Ibem et al. , 2013 ; Kalandides et al. , 2016 ; Parker, 2008 ; Pasquinelli et al. , 2018 ). The paper addresses a research gap related to the implementation of analytic tools in characterising place crowd dynamics identified by Helbing et al. (2007) , the paper provides further clarity in discerning what effective place management practice is, which is important because, as suggested by Kalandides et al. (2016) , improved knowledge of this can lead to the development of places that are successful, liveable and equitable. Finally, it can make a defined and measurable impact to place development, not just to specific crowd events but indeed to society as a whole.

This paper will review a selection of the available literature on crowd safety and crowd science in Section 2, followed by Section 3 outlining a crowd assessment safety risk model (the design, information, management-ingress, circulation, egress (DIM-ICE) risk model) used within this paper. Section 4 presents findings then reports on three case studies applying said model and a discussion engages the cases with appropriate literature. Finally, Section 5 describes conclusions and recommendations are made.

2. Literature review

2.1 crowd dynamics.

A crowd can be defined as follows: “a large group of individuals ( N ≥ 100 P) within the same space and at the same time whose movements continue for a prolonged period of time ( t ≥ 1960s) dependent on predominantly local interactions ( k ≥ 1 P/m 2 )” ( Duives et al. , 2013 , p. 194). It can be seen from this definition that the number N (number of individuals), k (density) and t (time) are the key elements of behaviour (movement/dynamics) of a crowd. However, a crowd is not simply a collection of a number of individuals; rather, it may exhibit highly complex dynamics.

lane formation;

oscillations at bottlenecks;

the faster-is-slower effect; and

clogging at exit ( Helbing et al. , 2000 , 2005 ; Kretz et al. , 2006 ; Georgoudas et al. , 2010 ; Shahhoseini et al. , 2018 ).

High crowd density may result in serious safety issues and crowd disasters. A frequent cause of crowd disasters is overcrowding ( Haase et al. , 2019 ). If congestion reaches critical levels, crowd motion transitions into a stop-and-go pattern and eventually causes a phenomenon called crowd turbulence ( Helbing et al. , 2007 ). Turbulent crowd motion is characterised by random, unintended displacements of groups in all possible directions (mass motion), which trigger disasters. Large gatherings of pedestrians are found in closed facilities such as shopping malls, stadiums, train stations and in open facilities such as walkways and parks. The number and severity of tragedy crowd disasters in high-density public events have risen significantly in the past decade ( Haase et al. , 2019 ). In a recent stampede during Hajj on 24 September 2015, more than 700 people died and more than 850 were injured. Therefore, understanding crowd behaviours and associated risks in high-density environments have the potential to save lives.

2.2 Crowd risk

Static crowd density and moving/dynamic crowd density have risks and different limits. Crowd risk level can be determined by integrating crowd density and flow rate as per Figure 1 ( Still, 2011 ). As shown in Figure 1 , density is measured by the number of people per m 2 and the flow rate is measured per metre per minute. Crowd risk increases with density and flow rate and moves into high risk when the density exceeds a certain point, e.g. five-person/m 2 suggested by ( Still, 2011 ). Daamen and Hoogendoorn (2003) outline the variance and created a higher flow rate standard in Holland, which was entered into Dutch law. Unfortunately, a “trained crowd” is not like a “tired crowd” leaving a venue or an “uncertain crowd” during an incident. The usefulness of Figure 1 is in defining the language of flow/density/risk, as the vocabulary is rather vague and the curve is difficult to visualise (without graphics). Using the descriptors free-flowing, capacity and congested provide a clearer definition of the “states” of the crowd flow.

these documents are biased towards overestimating risk;

there is often a “cut and paste” approach to the development of this type of documentation; and most importantly

there is often a lack of the information required to address relevant crowd safety issues.

Still (2014a , 2014b , p. 48) states that as follows:

the standard crowd risk analysis process of multiplying the likelihood of the risk occurring and the consequences of that risk fails the basic principles of Information Theory ( Shannon, 1948 ) in that it is impossible to reconstruct the conditions that give rise to many crowd related risks, especially those which are dynamic in nature.

Still’s (2013) research on crowd disasters found that the design element was the fundamental causality in over half of the incidents and concluded that an appropriate risk analysis of crowds needs to be undertaken to significantly reduce fatalities and serious injuries. This was also identified by Lak et al. (2019) and Ibem et al. (2013) , who suggested that appropriate design strategies, which consider participants’ preferences, improve the quality of places in different contexts and reduce safety risk. Similarly, Badiora and Odufuwa (2019) highlighted the importance of developing environmental designs for enhancing crowd safety. Fruin (2002) stated that there is a need for adequate formalised training in crowd management principles and techniques, to raise awareness and provide information regarding the appropriate tools that can be applied for an event to be organised successfully avoiding crowd safety risks.

2.3 Crowd science

crowd modelling;

crowd counting; and

crowd monitoring and management (including crowd risk analysis) ( Still , 2014a, 2014b ).

Each of these is discussed, in turn, below.

2.3.1 Crowd modelling.

Crowd modelling, based on simulating the crowd scenarios under various circumstances, is concerned with building robust representations of a crowd for scene understanding and is primarily a process used in the development of a robust crowd management system. The classical approaches for crowd modelling can be considered under macroscopic and microscopic scales ( Xu et al. , 2014 ; Bellomo et al. , 2016 ). Macroscopic models treat the crowd system as a whole and are usually designed to achieve real-time simulation for very large crowds, where each individual’s behaviour is not the primary research interest. In contrast, microscopic models are only for smaller crowds to achieve real-time simulation and focus on individual behaviours and their interactions based on complex cognitive models. Various simulation techniques have been developed such as situated cellular agents approach ( Bandini et al. , 2007 ). Virtual environment representations have also been constructed for crowd simulations ( Yersin et al. , 2005 ; Yersin et al. , 2009 ). A number of behaviour models have been proposed to investigate crowd behaviour ( Haghani and Sarvi, 2018 ) such as flow-based models and agent-based models. Crowd models may also incorporate different facets of a crowd. Some of the work targets the extraneous attributes of a crowd such as poses, movement patterns, appearance and coordinated positions of individuals; and some other work targets how crowds social behaviour emerges over time consequent to external events. However, the descriptive power of such methodologies for practical applications has remained unclear. This has been largely attributed to the lack of evidence-base against, which models can be calibrated or validated. In general, pure mathematical approaches or analytic models are not adequate in characterising crowd dynamics ( Helbing et al. , 2007 ).

2.3.2 Crowd counting.

Crowd counting is an important task for operational, safety and security purposes. Systems with these functions can be highly effective tools for crowd management ( Al-Zaydi et al. , 2016 ). Pedestrian crowd events are common but assessing the safety of such events has proven difficult. According to Duives et al. (2013) , not only are the layouts of the infrastructure different but also movements of pedestrians differ significantly between events. Different non-visual and visual methods are used for crowd counting and include various methods such as tally counters ( Lev et al. , 2008 ), differential weight counters ( dos Reis, 2014 ), infrared beams, wireless fidelity network and wireless sensor network based counters ( Yuan et al. , 2011 ; Di Domenico et al. , 2016 ). Visual-based crowd counting systems can be deployed using different types of cameras. According to Al-Zaydi et al. (2016) , methods based on computer vision are one of the best choices because cameras have become ubiquitous and their use increasing. There are an estimated six million CCTV cameras installed in the UK ( Birch et al. , 2017 ). In comparison with computer vision-based methodology, other non-visual methods need to be carefully planned and deployed for specific purposes and the accuracy is often less than a computer vision-based technology. Crowd counting based on computer vision can be classified into a line of interest and region of interest (Li et al. , 2011). Research into people counting in sparse environments is well established, but there are still many challenges and limitations to overcome in crowded environments ( Hou and Pang, 2010 ). They report on a lack of knowledge of how to handle occlusion (obstructions/blockages), which may only slightly affect crowd counting in sparse environments, but its effect increases significantly in crowded environments. Therefore, there is a need to develop a method to measure the level of occlusion, thereby improving the accuracy of counting. Experience demonstrates that event organisers typically inflate (grossly) their proposed numbers and the authorities then mobilise what they consider to be a proportionate response. The authorities turn out in force (expecting a larger crowd) and are seen as having provided a disproportionate response.

2.3.3 Crowd monitoring and management (including crowd risk analysis).

Crowd monitoring deals with constructing systems for real-time decision support through the statistical analysis of visual data. Furthermore, crowd management deals with the strategic, tactical and operational handling of crowds ensuring safety in an uncompromising yet efficient manner. Crowd risk analysis is considered to be an important aspect of crowd monitoring and management ( Smith, 2003 ). A number of studies highlight the shortcomings of the traditional approach for assessing risks at mass gatherings. First, a traditional risk assessment approach is insufficient to predict human behaviour ( Upton, 2004 ). Second, conventional risk assessment methods are biased towards overestimating risk ( Still, 2019 ). Despite the importance of the dynamic nature of crowd-related risks, the extant literature has not progressed the notion of monitoring, assessing or describing crowds to underpin interventions or controls. Also, there is limited evidence of practical applications for dynamic crowd risk analysis and monitoring.

2.4 The limits of current knowledge in the field

Fruin (1984) and Sime (1993) highlighted that there is a need to understand the interaction of efficient crowd management and place systems design for events, as these are the major factors that affect crowd disasters. Berlonghi (1995) argued that mismanagement of crowd risk may result in serious losses of life, health, property and money. Some operational guidelines and legislation are available (although crowd safety legislation and guidance are different across the UK, Europe, USA and Australia), which highlight measures that should be adopted in the context of crowd risk management regard the success of the delivery of events. For example, the “Green Guide”, in Guide to Safety at Sports Grounds ( Health and Safety Executive [HSE], 1997 ; GSSG, 2008 ), the “Primrose Guide” in Guide to Fire Precautions in Existing Places of Entertainment and Like Premises (Health and Safety Executive [HSE], 1998) and the “Purple Guide” in Event Safety Guide (Health and Safety Executive [HSE], 1999). The Safety at Sports Grounds 2018 Edition (Version 6) and the Primrose Guides (fire safety was replaced with a series of guides specific to premises) were replaced by the UK Building. The Purple Guide has also been extensively rewritten but is now subscription only.

However, there is still evidence that reveals clearly insufficient and inadequate planning at high-density public events ( Haase et al. , 2019 ; Shahhoseini et al. , 2018 ). Perhaps, these guidelines do not go far enough; the Fire codes set the times for evacuation, but this is based on the assumption of, namely, instant reaction to alarms; and the fire is no longer the only threat. We now have chemical, biological, radiological and nuclear threats, active shooter, etc. This means that the assumption of egress routes and viability, types of alarm, nature and direction of a threat, etc are all new variables and not the same as the “fire” assumptions.

3. Research methodology

Our study aims to provide empirical evidence of using relevant techniques for dynamic risk analysis to understand how to improve place crowd safety and throughput. To achieve this objective, the paper adopts a multiple case study approach ( Yin, 2018 ). However, gaining access to organisations and having permission to share the outcomes for this type of research can be difficult and is granted through a combination of good luck, effective planning and hard work ( Bell et al. , 2018 ). The paper has, therefore, presented a summary of projects that the lead author engaged with over time. The modelling tool presented in Table 1 , the “DIM-ICE risk model” was developed by Still (2009) from the application of evidence-based mathematical formulae and direct interactions with clients over an extended period. Primary data were collected based on the lead author’s experiences in these interactions. The case studies reported in the findings section, chosen to be representative of the scope and scale of place management within the word limits of a journal paper, contribute to the evidence base on the adoption and adaptation of place crowd safety and crowd science “best practice” within organisations.

information; and

management-related failures.

ingress (arrival);

circulation (movement within the venue); and

egress (departure).

The model combines these elements into a matrix framework ( Table 1 ). It is important to present this here to provide sufficient detail regards the different elements involved and to provide some contextualisation for the application of theory in the cases.

Table 2 presents the Crowd Safety Projects from 1999 to 2019 and provides a useful overview of the scope and scale of application of the techniques discussed in the paper. Importantly, it provides the evidence of the application of the DIM-ICE risk model to 68 projects and, when the singular public interactions per site (post-intervention) are counted, this equates to a total overall annual impact on 1,619,121,590 individuals who subsequently attended the venues.

Interestingly, presenting the data from Table 2 in a slightly different manner demonstrates that the projects split into three main groups ( Figure 2 ), according to the level of crowd density in the crowd safety projects from 1999 to 2019. Figure 2 shows the small, medium and large scale application of the DIM-ICE risk model across a number of representative places (e.g. religious festivals in the far east, multiple Olympic games, mass transit, etc.) characterised by crowd density. The following section reports on the actual application of the model at three large and challenging crowd safety projects as follows:

Sydney Olympics project;

Canary Wharf project;

Murrayfield Stadium project.

The case studies report on the use of the DIM-ICE risk model. The rationale of choosing these three projects was guided by our main research objective, to report on the implementation of an analytic tool in characterising place crowd dynamics. As discussed in the literature, there is a lack of clarity on the descriptive power of such methodologies for practical applications. Choosing three cases also provides us with opportunities to both expand this evidence base and to validate the DIM-ICE model.

things that go well;

things that need to be monitored; and

things that require improvement.

The DIM-ICE risk model also assisted in creating a one-way system to optimise crowd movement by using Network analysis theory ( Still, 2019 ). This one-way system included all possible options available to the attendees and introduced additional nodes. As part of it, a command and control tool was designed, which was an excel spreadsheet where information such as events start times, transit times, transit capacity, queuing times for security and screening, walk time to get to the stadium and the seating time, were all included. This approach assisted in identifying the number of people that would be in circulation (people moving between events) rather than in the stadiums. It provided an estimation of the expected number of pedestrians moving around the common domain. The method provided information to patrons that enabled them to manage their schedule to arrive at the stadium on time without experiencing unpleasant incidents such as long waiting times or an accident.

4. Findings

This section presents the findings from the application of the DIM-ICE risk model as Case 1, Case 2 and Case 3, respectively, in chronological order (from 2000 to 2018) to showcase the use of this model. Each of these cases outlines the context and impact of application; however, please note that some of the exact detail (especially regarding metrics and operational elements) have understandably had to be withheld for reasons of security.

4.1 Case 1 – Sydney Olympics project (2000)

The 24th Olympic games were held in September 2000 in Sydney, New South Wales, Australia. Approximately 11,000 athletes participated in the games and a total of 6.7 million games tickets were sold (92.4 per cent of the availability). Before the Sydney 2000 Games started, the organisers realised that they had to deal with a significant challenge as follows: how to design optimal throughput, minimise the exposure to crowd safety risk and provide a successful and pleasant event. Crucially, they had to deal with non-event ticketing spectators (people that just wanted to be part of the Olympics atmosphere). Although they knew the capacity of the domain, they could not estimate the number of non-event ticketing spectators attending the games. Therefore, they requested external advice on how to face this issue because they did not have the knowledge and skills in-house. Table 3 presents a summary of the Sydney Olympic Games project.

The DIM-ICE risk model was applied to the circulation and movement of the people attending the event(s). The implementation of the model aimed to identify areas that required improvement (e.g. expected high-density) and reduce the risk of crowd safety-related issues. Figure 3 presents the DIM-ICE risk model created for the Sydney Olympics project. Specifically, as Figure 3 summarises, during normal conditions, the main challenges that the model identified were as follows:

Design category: difficulties in estimating the number of non-event ticketing spectators, which caused a risk of overcrowding during the ingress, circulation and egress phases.

Management category: difficulties in managing peoples’ movements, especially from North-South in the event area during the ingress and circulation phases.

Under an emergency condition, the DIM-ICE risk model assisted in realising that there were a number of potential crowd safety-related issues that the organisers had to address to offer a safe environment.

Design category: during the circulation phase, there was a limited flow capacity. Considering that the number of non-event ticketing spectators could not be estimated, this may have caused difficulties in attendees’ movements and safety issues.

Information category: there was unclear signage during the circulation and egress phases.

Management category: the model identified that a detailed evacuation plan was missing, another potential crowd safety issue.

The data analysis revealed that, approximately, it would take at least 2 h to get from places like Sydney into the seating area. However, the attendees were informed, by the organisers of the Sydney 2000 Games that it would take 4 h to arrive at the venue. The reason, to avoid crowd safety-related issues and provide a greater distribution of people’s arrival times. Patrons were satisfied because there was a variety of events/entertainment in the concessions around the stadiums in the common domain spaces; plus, they arrived in the venue on time/in plenty of time (under promise and over deliver).

Based on the data analysis, an estimation regarding the demand at the maximum point of the system was achieved. In total, 80,000 people per hour was the full capacity of the system – this was calculated by optimising routes, making them wide enough (e.g. 20 m wide to facilitate a flow rate of 82 people per metre per minute), etc. As a result, the DIM-ICE risk model helped to designed resilience into the system and ensured that the peak demand represented 80 per cent of the system’s capacity consistently allowing greater efficiency.

4.2 Case 2 – Canary Wharf project (2003)

In the early 2000s Canary Wharf in London, UK (a major finance hub) experienced at least two “bomb threats” per week ( Mullin et al. , 1996 ). Reports of vehicles laden with explosives or the existence of so-called “parcels bombs” in particular places were a regular occurrence. The UK’s London Metropolitan Police Service was facing complications in dealing with these issues because it did not have an official or legal authority to call for an evacuation of areas or premises unless there was an imminent risk to life (there are both legal and financial restrictions to evacuating a major banking centre due to the high number of false alarms). The Metropolitan Police Service could not force an evacuation because of the potential for false alarms and there was no system in place to quantify the level of threat. Therefore, it was decided that a top-down command and control evidenced-based procedure was required for any building to be evacuated; this was serving the dual purpose of protecting lives and minimising the potential risk of litigation.

The unique positioning of Canary Wharf London is complicated because it is an island surrounded by water, it, therefore, has limited exit points and emergency access routes. The challenge of providing directional information (i.e. which was the safe way to go) to people to move away from the potential threats become complex. Table 4 provides an overview of the Canary Wharf project.

Applying the DIM-ICE risk model, a transparent and robust network analysis was conducted. This served to analyse the capacity of available roads and optimised the crowd moved away from the location of the threat. Figure 4 presents the DIM-ICE risk model applied for the Canary Wharf project. The areas that required further improvements for any potential crowd safety issues to be avoided were identified. As Figure 4 summarises, under normal conditions, the main challenges were as follows:

Design category: the limited capacity of available roads, which could cause serious safety issues as there were difficulties in directing people to move away from potential threats during the ingress, circulation and egress phases.

Management category: the Metropolitan Police Service was facing difficulties in managing information regarding the capacity and availability of routes during the ingress phase.

Under emergency conditions, the DIM-ICE risk model identified similar potential crowd safety-related issues.

Design category: the limited capacity of available roads negatively impacted the design of an optimum route to keep people away from potential threats.

Information category: a significant issue was the lack of information to identify the severity of the threat and on routes that had the maximum capacity to direct people to move away from potential threats.

Management category: the model identified the need for a detailed evacuation plan, to help the Metropolitan Police Service follow standardised procedures, avoiding potential issues related to crowd safety.

Based on the DIM-ICE risk model, bespoke software was developed by the London Metropolitan Police Service for Canary Wharf, representing the entire site using coordinated grids. The location of any potential threat was indicated by a specific grid reference and the severity of the threat identified by a physical radius. The algorithms within the software provided the information required to identify routes that had the maximum capacity to direct people and move them away as safely as possible. A complicating factor was that the accessibility of routes was constantly changing (building works, repairs, road layout changes, etc.), which affected the capacity of specific routes. Therefore, the software required constant updating and development for significant information regarding the capacity and availability of routes to be as robust and accurate as possible. The use of the DIM-ICE risk model provided the much-needed evidence-based solution to a highly complex and changeable problem area.

The project had three core modelling element, a generic routing diagram (as a lookup table) and a specific threat location exclusion network (only showing viable routes depending on location and severity of the threat). This was coupled to information-based maps and video clips for training the occupants and a “pied piper approach” to egress. That is, instead of pushing people out of buildings, marshals were trained to lead people away from the threat. The instructions for the occupants were to “follow the crowd and keep going”. In essence, this meant there developed a directional approach to all possible threat locations. This was coupled to a training programme and situation awareness for all levels of command and control for the site.

4.3 Case 3 – Murrayfield stadium project (2018)

Murrayfield Stadium is a sports stadium located west of Edinburgh, the capital of Scotland. It is the largest and most impressive stadium in Scotland and the fifth largest in the UK with a seated capacity of 67,144. It is the home of Scottish rugby and Murrayfield Stadium has also hosts musical events.

In November 2018, expert advice on crowd safety was required for two reasons as follows: a new building was to be developed, close to the site, which would impact upon the layout of the merchandising area both during and after its construction. The physical change to the site were known, but the impact during construction and on the merchandising locations post-build were under negotiation. An investigation into how the crowd was going to adapt to the new environment was required; there were two minor injuries on-site, which made the operations team of the Murrayfield stadium seek advice on whether existing crowd safety procedures were sufficient from both a health and safety perspective and potential exposure to litigation. The use of the DIM-ICE risk model assisted in analysing the current situation and providing a set of recommendations. Table 5 summarises the Murrayfield Stadium project.

To develop the DIM-ICE risk model, data was collected through conducting a site survey (a day on the site as a customer – walking to various areas, setting up cameras, extensive videoing of ingress, circulation and egress); using the data from the stadium CCTV cameras and also reviewing the existing risk assessment policies. Based on the data analysis, the DIM-ICE matrix was created, which highlighted areas that required improvements. Figure 5 presents the DIM-ICE risk model applied to the Murrayfield Stadium project. More than 90 per cent of Murrayfield Stadium’s safety operations were well above the standards appropriate for stadia of this size, but there were critical elements that required review. Areas that required improvement were identified and the design of the queuing system improved to achieve more efficient inflow. The analysis of the data demonstrated that there was a significant risk on the site – a narrow area at the west end of the stadium with the rather high-density flow (as a result merchandising in this area was highly profitable). There were also waste bins in that area, which further reduced the width. Specifically, as Figure 5 summarises, the main areas that required involvements were related to the existence of an emergency situation as follows:

Design category: the identified narrow area could cause overcrowding at the ingress, circulation and egress phases.

Information category: visitors might have difficulties in exiting the stadium due to unclear signage, which could cause safety issues.

Management category: the model identified that the risk assessment processes had to be updated, which was a critical requirement for creating a safe environment.

In addition, the development of the new building, by changing dynamics of flow, changed the routing and footfall moving in and out the stadium, specifically if an emergency situation required a directional egress close to the construction site during the build. A number of recommendations were made, aiming to provide solutions to the identified safety issues. Initially, the operations team were recommended to: update the risk assessment processes, to increase the width of the west end area with the tight footfall (density); and to move the waste bins units out of the high footfall (volume) area, which would reduce the flow constraint to make it more efficient and accessible for patrons. Finally, an engagement between the operations team and the building construction engineers was recommended to ensure that specific corridors would remain clear and available for the crowd, during the construction phase. Note: one of the modelling tools used here was the risk/congestion mapping, a visual approach to identifying areas of high footfall and potential congestion at specific times during an event (ingress/circulation/egress).

5. Discussion and final remarks

To add value to the discussion of safe place management and make a defined contribution the paper has reported on the use and intervention of a crowd assessment safety risk model, the “DIM-ICE risk model”. The aim of the study was to report on the impact and use of this intervention on place crowd safety and the application of crowd science. In addition, we offer the following learning points for practitioners, drawing upon the significant background provided by the presented cases.

5.1 What do we learn from the application of the design, information, management-ingress, circulation, egress model?

The model was developed from the analysis of past disasters and their fundamental causes. So the model helps users consider the phases and influences of crowd behaviour through the lens/filter of causality. In essence, it draws the attention of the user to think through the event in time (ICE) and controls (DIM). In response to these issues, the DIM-ICE risk model was used to compare operational situations during so-called “normal” and then “emergency” conditions for each phase and aspect of proceedings to fully scope out crowd dynamics at an event. Helbing et al. (2007) clearly state that that pure mathematical approaches and analytic models are not adequate for this purpose, that the conventional risk assessment process is mathematically biased.

5.2 What works, in what contexts?

well organised (green);

those that require improvement or monitoring during an event (amber); and

those that must be improved (red).

Haase et al. (2019) reported on overcrowding being a frequent cause of crowd disasters, this highly visual method serves to clearly pinpoint where and potentially when this might occur. When applied and facilitated by an informed cross-discipline operational team the approach allows the potential rapid identification of high-risk areas. Alternative place design, communication and management requirements can then be identified, discussed, considered and implemented to allow the development and adoption of appropriate operational strategies on site. This goes beyond the rather rudimentary aspects outlined in the available literature by Helbing et al. (2000) , Helbing et al. (2005) , Kretz et al. (2006) , Georgoudas et al. (2010) and Shahhoseini et al. (2018) , e.g. lane formation, oscillations at bottlenecks, the faster-is-slower effect and so-called “clogging” at exits. This contribution is important as Haase et al. (2019) described the number and severity of fatal crowd disasters in high-density public events rising significantly.

As per Duives et al. (2013) , adequate crowd movement information (facilities, locations, correct exits in an emergency, etc.) must be clearly communicated to the attendees/public in a manner that removes any scope for ambiguity. The DIM-ICE risk model, therefore, has the potential to provide an evidence-based comprehensive pre-event analysis, which properly forecasts potential crowd issues ensuring a successful event, where the attendees enjoy a positive experience ( Al-Zaydi et al. , 2016 ). Upton (2008) suggested that successful planning of an event requires an in-depth risk assessment of crowd safety, which the reported intervention of the DIM-ICE risk model achieves. The DIM-ICE risk model can be used across the spectrum – to show green/amber/red in context and as a methodology for systematic and rigorous assessment of the risks using phases of behaviour and influences on behaviour through the design of the place, information and management. In essence, it provides a “how to shape the crowds' behaviour” guide.

5.3 What does not work?

The DIM-ICE risk model is risk/causality-based. It does not define routing-area-movement-people. However, when used with other tools and knowledge it is a comprehensive risk-based analysis method. The implementation of the model assists event planners to identify locations and potentially times of high risk and control them, through effective place design, information systems and management strategies. Still (2015a , 2015b ) suggested that this model provides solutions to complex safety issues by simplifying them into elements for consideration, a multiscale approach to the overall process of event planning (zoom into a section, then zoom out to see the overall impact on the event), providing the information required for the planning phase. It sits between the traditional approaches of crowd modelling, considered under macroscopic (treat the crowd system as a whole) and microscopic (smaller crowds, real-time simulation, individual behaviour and interactions) scales ( Xu et al. , 2014 ; Bellomo et al. , 2016 ) and the impressive but resource-intensive virtual environment representations ( Yersin et al. , 2005 ; Yersin et al. , 2009 ). In a similar manner to the above, the application of the DIM-ICE risk model is required prior to an event to establish a crowd risk plan; for example, ingress and egress routes have to be of sufficient size to safely accommodate predicted crowds, therefore, avoiding congestion that may occur. However, it is the robustness and relevance of its application that appears to make a difference ( Still , 2014a, 2014b ) creating a much-needed evidence base ( Helbing et al. , 2007 ) to inform operational decision making.

5.4 What limitations this model presents?

It is only one of a number of tools that can/should be used regards safety in place management. Together they cover the risk dynamics. So it is part of a suite of tools – specific to each place. The case studies presented within this paper seek to inform the challenges and limitations of crowded environments that event organisers face, as highlighted by Hou and Pang (2010) . The paper provides information regarding the application of an evidence-based approach (the DIM-ICE risk model), addressing the research gap related to the implementation of analytic tools in characterising crowd dynamics identified by Helbing et al. (2007) . It contributes to what Chai et al. (2017) identified as the emerging field of research motivated by crowd safety issues in our increasingly congested environments. It also provides further understanding to place management terms as “improved knowledge and more effective place management practice can ultimately lead to places that are more successful, more liveable and more equitable” ( Kalandides et al. , 2016 , p. 358).

To take this development further we suggest more research into aspects of crowd safety, crowd science and its application. Ultimately this goes beyond simple reporting and has the very great potential to make a defined and measurable impact, not just to specific crowd events but also to society as a whole.

Crowd density vs crowd flow rate

Crowd density and crowd safety projects taken place from 1999 to 2019

The DIM-ICE risk model – the Sydney Olympics project

The DIM-ICE risk model – the Canary Wharf project

The DIM-ICE risk model – the Murrayfield Stadium project

The DIM-ICE risk model

	Ingress	Circulation	Egress

Design	Elements of the design that influence the crowd during ingress – this specifically relates to the elements of the design (such as barriers, local geometry, width of routes, paths and stairs, entrances, turnstiles, etc.	Elements of the design that influence the crowd during circulation (this relates to “mid-event” – moving around) such as route widths, stairs, layout and facilities management, concessions, etc.	Elements of the design that influence the crowd during egress (getting out) – specifically the egress capacity, route complexity and geometry (stairs, corridors, doors, gates, etc.)
Information	Prior to the event, many things can influence crowd behaviour such as advanced notifications, media coverage, tickets and posters, local knowledge, previous event history, nature of the band, weather forecasts. Assess how the information prior to the event, near the event, on the way to the event and at the venue could influence the crowd – specifically signage and information systems	Mid-event there could be a lot of conflicting information, the performance, the concessions, signage, PA announcements, stewards and information points. Assess how this influences the crowds and how best to inform the crowd of the facilities	Signage and PA announcements for departure (non-emergencies) influence not only the direction but the distribution of the crowd. Ensure that all routes are clearly signed – checking for lines of sight to ensure all exit routes are visible
Management	Stewards, security and police management not only divert the crowd to the most appropriate areas but also influence the crowd’s behaviour (such as reducing the element of hooliganism by increasing the visibility of police – this is also information). Queues can be actively managed and evenly distributed if approach routes allow good sightlines	During the event, the stewards can actively manage queues and crowd movements	During egress departing crowds can be actively managed = specifically car parks can be made more efficient if actively managed (rather than allowing a free-for-all dash for the exit)

Design	How does the ingress system cope during an emergency – you may need to consider a “stay out” strategy and assess how the design copes with turning the crowd back from an internal threat	Mid-event how quickly can this site evacuate – typically the type of calculation a fire/safety officer would perform to ensure the site had sufficient egress routes and capacity for clearance	How does the egress system cope during an emergency – you may need to consider a “stay put” strategy and assess how the design copes withholding the crowd back from an external threat
Information	During ingress how would the crowds be informed of an emergency? What type of information, in what form and content is required?	Mid-event how would the crowds be informed of an emergency? What type of information, in what form and what content is required? Ensuring the crowd moves away from the threat requires more than just a please leave an announcement	During egress how would the crowds be informed of an emergency? What type of information, in what form and content is required? For this, you need to consider the crowd in the process of normal egress
Management	During ingress, there may be more people trying to gain entry than is physically possible (for example, a “free” event). The crowds may need active management to prevent overcrowding in the event space. This would be considered an emergency situation as there is a risk of crushing if the event does not have an active management system	During the event, the crowd may need to be managed (directed) away from a threat. Consider the information (above) and the management of the egress for a direction that ensures the crowd moves as quickly as possible away from the source of danger	The crowd may need to be managed after vacuation (say on a holding area) to be kept safe until the threat/danger has passed You may need to keep managing the crowd for several hours during a holding operation. You will need to keep the crowd informed until it is safe to let the crowd disperse

Type of event	Sports events	Religious festivals	Major events	City centre and retail	Festivals and street events	Rail	Leisure	Evacuation
	Sydney Olympics Beijing Olympic Stadium London Olympics Twickenham (UK) Swedbank Arena Project (Sweden) Commonwealth Games (UK) Millennium Stadium (UK) Wembley Stadium (UK) Football Licensing Authority Concourses (UK) Hong Kong Jockey Club Premier League persistent standing (modelling project) Penn State (USA) Everton Fan Zone (UK) Etihad – Manchester City Stadium (UK) Beşiktaş – Vodafone Arena (Turkey) Liverpool Stadium Murray field Stadium	Jamarat Bridge, Saudi Arabia Al-Haram, Saudi Arabia Al Mashaaer AlMugaddassah metro project bid Saudi Arabia Diwali (Leicester)	Royal wedding MCFC and MUFC victory parades, Manchester Leicester Caribbean Carnival Glastonbury music festival, UK Hampton Court flower show Great Manchester run London New Year Event Aberdeen Hogmanay Mathew Street Festival (Liverpool) Penn Stadium (Nebraska v Penn State) Lincoln Christmas Market Birmingham Christmas Market Kendal Torchlight Parade T in the Park (Scotland) Canada Day (Ottawa) Dubai New Year	Paradise Street Development Area, Liverpool Jabal Omar development (Makkah – Saudi Arabia) West Kowloon Cultural District bid Covent Garden Ticket Hall Retail Analysis Austin, Texas (South by Southwest Music Festival) Manchester Arena EMAAR – Dubai Mall – UAE Toronto (Metrolinx)	Chelsea flower show Lewes fireworks Aberdeen fireworks Royal Parks (London) Protest March (Manchester – police planning)	Cubic Transportation Ltd (London Underground) Easingwold training courses (London Underground Limited and British Transport Police), Dwell modelling – Alstom, Porterbrook, Bombardier, Interfleet Wembley Complex Station (Marshalling Study) Chesterfield Railway Station Covent Garden Ticket Hall Retail Analysis	Kuala Lumpur Conference Centre Aquarium Cleethorpes Outdoor Arena Development Manchester Museum National Arenas Association Liverpool Arena and Conference Centre Premier League (UK football) – safety in stands (modelling project)	Canary Wharf (London Financial District) Labour Party Conferences Amsterdam Police Barclays Bank AWE Aldermaston (public enquiry), Bluewater shopping mall and Westfield (Olympics gateway)
Sum of projects	17	4	16	8	4	6	6	7
Annual number of people who then attend the places	39,599,850	10,000,120	9,907,000	174,169,000	1,270,000	1,304,360,500	3,050,100	76,765,020

The summary of the Sydney Olympics project

Project	Sydney Olympics project, Sydney, Australia
When	2000
Aim	Optimising the crowd throughput Minimising the exposure to crowd safety risk Providing a successful and pleasant event
Problem	A new building was developing – would the crowd adapt to the new environment? There were two minor injuries on site – were the existing safety procedures adequate?
Solution	The DIM-ICE risk model applied emphasising more on the circulation movement of the people attending the event The development of a one-way system optimising crowd movement by using the Network analysis theory
Benefits	The identification of the expected number of pedestrians moving in the common domain Information provided to patrons that enabled them to manage their schedule to arrive to the stadium on time without experiencing any unpleasant incident The design of the queuing system was improved to achieve more efficient inflow A successful event was achieved Customer satisfaction was enhanced
People impacted by the project recommendations	6,800,000 during the games

The summary of the Canary Wharf project

Project	Canary Wharf project, London, UK
When	2003
Aim	Optimising the crowd throughput
Problem	Canary Wharf was experienced an incredible crowd safety restated tread and the police did not have the official or legal authority to call an evacuation
Solution	The use of DIM-ICE risk model assisted in analysing the capacity of all the existing available roads and optimising the crowd movement away from the location of the thread
Benefits	Information provided regarding the route that had the maximum capacity to direct people and move them away from the thread
People impacted by the project recommendations	90,000 per day

White squares = areas that are sufficient, no improvements are required; light grey squares = require improvement; and grey squares = require much improvement to avoid crowd safety issues.

Project	Murrayfield Stadium, Edinburgh, UK
When	2018
Aim	Optimising the crowd throughput
Problem	A new building was developing – would the crowd adapt to the new environment? There were two minor injuries on site – were the existing safety procedures adequate?
Solution	The use of DIM-ICE risk model assisted in analysing the current situation and providing a set of recommendation
Benefits	>90% of Murrayfield Stadium’s safety operations were appropriate Areas that required improvement regarding the existing risk assessment policy were identified The design of the queuing system was improved to achieve the more efficient inflow Murrayfield Stadium demonstrate continual assessment and improvement to the risk management process Customer satisfaction was enhanced
People impacted by the project recommendations	67,000 per event

Al Jazeera ( 2019 ) Thousands stage global climate protests before UN summit, Al Jazeera and News Agencies, available at: https://aljazeera.com/news/2019/11/thousands-stage-global-climate-protests-summit-191129140130762.htm

Al-Zaydi , Z.Q. , Ndzi , D.L. , Yang , Y. and Kamarudin , M.L. ( 2016 ), “ An adaptive people counting system with dynamic features selection and occlusion handling ”, Journal of Visual Communication and Image Representation , Vol. 39 , pp. 218 - 225 .

Badiora , A.I. and Odufuwa , B.O. ( 2019 ), “ Fear dynamics in public places: a case study of urban shopping centers ”, Journal of Place Management and Development , Vol. 12 No. 2 , pp. 248 - 270 .

Bandini , S. , Federici , M.L. and Vizzari , G. ( 2007 ), “ Situated cellular agents approach to crowd modeling and simulation ”, Cybernetics and Systems , Vol. 38 No. 7 , pp. 729 - 753 .

Bell , E. , Bryman , A. and Harley , B. ( 2018 ), Business Research Methods , Oxford University Press , Oxford .

Bellomo , N. , Clarke , D. , Gibelli , L. , Townsend , P. and Vreugdenhil , B. ( 2016 ), “ Human behaviours in evacuation crowd dynamics: from modelling to “big data” toward crisis management ”, Physics of Life Reviews , Vol. 18 , pp. 1 - 21 .

Berlonghi , A.E. ( 1995 ), “ Understanding and planning for different spectator crowds ”, Safety Science , Vol. 18 No. 4 , pp. 239 - 247 .

Birch , P. Young , R. and Chatwin , C. ( 2017 ), “ Smart cameras? II ”.

Chai , Y. , Miao , C. , Sun , B. , Zheng , Y. and Li , Q. ( 2017 ), “ Crowd science and engineering: concept and research framework ”, International Journal of Crowd Science , Vol. 1 No. 1 , pp. 2 - 8 .

Daamen , W. and Hoogendoorn , S.P. ( 2003 ), “ Experimental research of pedestrian walking behavior ”, Transportation Research Record: Journal of the Transportation Research Board , Vol. 1828 No. 1 , pp. 20 - 30 .

Di Domenico , S. , Pecoraro , G. , Cianca , E. and De Sanctis , M. ( 2016 ), “ Trained-once device-free crowd counting and occupancy estimation using wifi: a doppler spectrum based approached ”, IEEE 12th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob) , IEEE , pp. 1 - 8 .

Dos Reis , J.V.D. ( 2014 ), “ Image descriptors for counting people with uncalibrated cameras ”.

Duives , D.C. , Daamen , W. and Hoogendoorn , S.P. ( 2013 ), “ State-of-the-art crowd motion simulation models ”, Transportation Research Part C: Emerging Technologies , Vol. 37 , pp. 193 - 209 .

Fruin , J. ( 1984 ), “ Crowd dynamics and auditorium management. In an international association of auditorium managers ”, Auditorium News , available at: www.gkstill.com/Support/crowd-flow/fruin/Fruin3.html

Fruin , J.J. ( 2002 ), “ The causes and prevention of crowd disasters ”, 20 January , available at: www.crowdsafe.com/fruincauses.pdf

Georgoudas , I.G. , Sirakoulis , G.C. and Andreadis , I.T. ( 2010 ), “ An anticipative crowd management system preventing clogging in exits during pedestrian evacuation processes ”, IEEE Systems Journal , Vol. 5 No. 1 , pp. 129 - 141 .

GSSG ( 2008 ) Guide to Safety at Sports Grounds, Fifth Edition, Department for Culture, Media and Sport, Her Majesty’s Stationery Office.

Haase , K. , Kasper , M. , Koch , M. and Müller , S. ( 2019 ), “ A pilgrim scheduling approach to increase safety during the Hajj ”, Operations Research , Vol. 67 No. 2 , pp. 376 - 406 .

Haghani , M. and Sarvi , M. ( 2018 ), “ Crowd behaviour and motion: empirical methods ”, Transportation Research Part B: Methodological , Vol. 107 , pp. 253 - 294 .

Health and Safety Executive [HSE] ( 1997 ), “ Fairgrounds and amusement parks: guidance on safe practice ”, HSE Publication HS(G) 175 .

Health and Safety Executive [HSE] ( 1998 ), Guide to Fire Precautions in Existing Places of Entertainment and like Premises , HMSO London .

Health and Safety Executive [HSE] ( 1999 ), “ The event safety guide: a guide to health ”, Safety and Welfare at Music and Similar Events , 2nd ed. Surrey: HSE Books .

Helbing , D. , Farkas , I. and Vicsek , T. ( 2000 ), “ Simulating dynamical features of escape panic ”, Nature , Vol. 407 No. 6803 , pp. 487 .

Helbing , D. , Johansson , A. and Al-Abideen , H.Z. ( 2007 ), “ Dynamics of crowd disasters: an empirical study ”, Physical Review E , Vol. 75 No. 4 , pp. 046109 .

Helbing , D. , Buzna , L. , Johansson , A. and Werner , T. ( 2005 ), “ Self-organized pedestrian crowd dynamics: experiments, simulations, and design solutions ”, Transportation Science , Vol. 39 No. 1 , pp. 1 - 24 .

Hou , Y.L. and Pang , G.K. ( 2010 ), “ People counting and human detection in a challenging situation ”, IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and Humans , Vol. 41 No. 1 , pp. 24 - 33 .

Ibem , E.O. , Uwakonye , O. and Aduwo , E.B. ( 2013 ), “ An appraisal of urban renewal in Nigeria a case study of the Nigerian army shopping Arena, Oshodi-Lagos ”, Journal of Place Management and Development , Vol. 6 No. 2 , pp. 155 - 170 .

Kalandides , A. , Millington , S. , Parker , C. and Quin , S. ( 2016 ), “ Shopping districts and centres, markets, neighbourhoods, public squares, and urban gardens reflecting upon place management practice in Berlin ”, Journal of Place Management and Development , Vol. 9 No. 3 , pp. 351 - 359 .

Kretz , T. , Wolki , M. and Schreckenberg , M. ( 2006 ), “ Characterizing correlations of flow oscillations at bottlenecks ”, Journal of Statistical Mechanics: Theory and Experiment , Vol. 2006 No No. 02 , p. P02005 .

Lak , A. , Ramezani , M. and Aghamolaei , R. ( 2019 ), “ Reviving the lost spaces under urban highways and bridges: an empirical study ”, Journal of Place Management and Development , Vol. 12 No. 4 , p. 469 .

Lev , L. Brewer , L.J. and Stephenson , G.O. ( 2008 ), “ Tools for rapid market assessments ”,

Mullin , J. Campbell , D. Wintour , P. Sharrock , D. Chauchary , V. Pallister , D. and Nowicka , H. ( 1996 ), “ IRA smash ceasefire ”, The Guardian, Sat 10 Feb 1996 (retrieved 26/09/19) .

Parker , C. ( 2008 ), “ Extended editorial: place – the trial frontier ”, Journal of Place Management and Development , Vol. 1 No. 1 , pp. 5 - 14 .

Pasquinelli , C. , Koukoufikis , G. and Gogishvili , D. ( 2018 ), “ Beyond eventification: capacity building in post-disaster temporariness ”, Journal of Place Management and Development , Vol. 12 No. 1 , pp. 88 - 101 .

Shahhoseini , Z. , Sarvi , M. and Saberi , M. ( 2018 ), “ Pedestrian crowd dynamics in merging sections: Revisiting the “faster-is-slower” phenomenon ”, Physica A: Statistical Mechanics and Its Applications , Vol. 491 , pp. 101 - 111 .

Shannon , C.E. ( 1948 ), “ A mathematical theory of communication ”, Bell System Technical Journal , Vol. 27 No. 3 , pp. 379 - 423 . 623-656, July, October .

Sime , J.D. ( 1993 ), “ Crowd psychology and engineering: designing for people or ballbearings ?”, in Smith , R.S. and Dickie , J.F. (Eds), Engineering for Crowd Safety , Elsevier Science Publishers: BV .

Smith , K. ( 2003 ), Environmental Hazards: assessing Risk and Reducing Disaster , Routledge .

Still , G.K. ( 2000 ), Crowd Dynamics , University of Warwick .

Still , K. ( 2009 ), “ Safety in numbers ”, ISquared, Summer , pp. 23 - 26 .

Still , G.K. ( 2011 ), “ Standing crowd density ”, available at: www.gkstill.com/Support/crowd-density/CrowdDensity-1.html

Still , G.K. ( 2014a ), Introduction to Crowd Science , CRC Press .

Still , K. ( 2014b ), “ Visualising risk assessment for crowd safety ”, Journal of the International Centre for Sports Security , Vol. 2 No. 1 , available at: http://icss-journal.newsdeskmedia.com/images/Upload/Vol_2_n0_1/ICSS_Vol2.1.pdf

Still , K. ( 2015a ), “ Fruin – levels of service ”, available at: www.gkstill.com/Support/crowd-flow/fruin/Fruin1.html

Still , K. ( 2015b ), Crowd Science and Risk Analysis. In Level 5 Diploma in Crowd Science , 25-27 May 2015 , Brisbane .

Still , G.K. ( 2019 ), “ Crowd science and crowd counting ”, Impact , Vol. 2019 No. 1 , pp. 19 - 23 .

Upton , M. ( 2004 ), “ Risk analysis for major concert events, the benefits of hindsighted ”, Cabinet Office Seminar on Safety at Mass Crowd Events (January 2004) .

Upton , M. ( 2008 ), “ Safe event management ”, Paper delivered to the Theatre Managers Association Conference , Birmingham, 10 June 2008.

Xu , M.L. , Jiang , H. , Jin , X.G. and Deng , Z. ( 2014 ), “ Crowd simulation and its applications: recent advances ”, Journal of Computer Science and Technology , Vol. 29 No. 5 , pp. 799 - 811 .

Yersin , B. , Maïm , J. , Pettré , J. and Thalmann , D. ( 2009 ), “ Crowd patches: populating large-scale virtual environments for real-time applicationsed ”, Proceedings of the 2009 symposium on Interactive 3D graphics and games , ACM , pp. 207 - 214 .

Yersin , B. , Maïm , J. , Ciechomski , P. , Schertenleib , S. and Thalmann , D. ( 2005 ), “ Steering a virtual crowd based on a semantically augmented navigation graphed ”, Proc. The First International Workshop on Crowd Simulation (V-CROWDS’05) , Lausanne , Switzerland , pp. 169 - 178 .

Yin , R.K. ( 2018 ), Case Study Research and Applications , SAGE Publications London .

Yuan , Y. , Qiu , C. , Xi , W. and Zhao , J. ( 2011 ), “ Crowd density estimation using wireless sensor networksed ”, Seventh international conference on mobile Ad-hoc and sensor networks , IEEE , 138 - 145 .

Corresponding author

Related articles, all feedback is valuable.

Please share your general feedback

Report an issue or find answers to frequently asked questions

Contact Customer Support

Crowd Management – Navigating Challenges and Implementing Best Practices in Crowd Management

Posted: 5 Dec 2023

Namreen Asif

Srinivas University

Date Written: November 30, 2023

Since more and more events are drawing large crowds, crowd management has become a crucial subject. The abstract explores the intricate area of crowd control, looking at the challenges faced by event organizers and the evolving tactics employed to ensure both public safety and the greatest possible experience for guests. A deep understanding of the challenges and the use of innovative solutions are crucial in the dynamic field of crowd management. Planning a large-scale event effectively necessitates an awareness of how technology, security, and human behavior interact. As events expand in size and complexity, crowd management strategies must be continuously improved to ensure the safety, security, and enjoyment of participants. With this abstract, a thorough analysis of the intricacies and advancements in the subject of crowd management might be started. The number of people attending public meetings has increased, and metropolitan crowd densities are continuing to climb, making crowd control more challenging than ever. Each year, many people die as a result of poor crowd control and planning. Crowd management is an interdisciplinary discipline that requires a grasp of technical and technological issues since crowd behavior and flow management entail both psychological and social elements. A wide range of crowd management best practices and challenges for effective systems are covered in this article.

Keywords: Event Planning, Sustainable Design, Crowd Management, Continuous Innovation,

JEL Classification: IEC23213

Suggested Citation: Suggested Citation

Namreen Asif (Contact Author)

Srinivas university ( email ).

College of Management and Commerce, Mangalore Mangalore, 575001 India 9740514002 (Phone)

Do you have a job opening that you would like to promote on SSRN?

Paper statistics, related ejournals, sustainable technology ejournal.

Subscribe to this fee journal for more curated articles on this topic

Find anything you save across the site in your account

Crush Point

On Thanksgiving Day, 2008, shoppers began lining up outside the Wal-Mart in Valley Stream, Long Island, at 5:30 P.M ., near a small, handwritten sign that read “Blitz Line Starts Here.” Like many other retailers holding “doorbuster” Black Friday sales, Wal-Mart was offering deep discounts on a limited number of TVs, iPods, DVD players, and other coveted products. Only two months earlier, the U.S. economy had nearly collapsed, and although the Christmas shopping season was looking dismal, there was still some dim hope that the nation might be able to shop its way out of disaster, as we were advised to do after 9/11.

By two in the morning, the line ran the length of the building, past Petland, turned at a wire fence, and stretched far into the bleak parking lot of the Green Acres Mall, a tundra of frosted tarmac. There were already more than a thousand people. Store managers had placed eight interlocking plastic barriers between the front of the line and the outer doors to the store, to create a buffer zone that would keep people from crowding around the entrance. But at three people began jumping the barriers. The store’s assistant manager, Mike Sicuranza, spoke to the manager, Steve Sooknanan, who had gone to a hotel to rest, and told him that customers had breached the buffer zone. Sicuranza sounded frightened. Sooknanan told him to call the police.

The Nassau County police arrived soon after the call, and, using bullhorns, ordered everyone to get back behind the barriers. The police were still there at four, when Sooknanan returned to the store. Shortly afterward, a Wal-Mart employee brought some family members inside the barriers, angering the crowd. About two hundred shoppers pushed into the buffer zone. Those in front were squeezed against two sliding glass outer doors that led into a glassed-in, high-ceilinged entrance vestibule that also held some vending machines. These had been pushed to the center of the space, to prevent people from crossing it diagonally and entering through the exit doors. As more people gathered, in anticipation of the store’s opening, at 5 A.M. , the pressure on the doors built and they began to shake. “Push the doors in!” some chanted from the back.

Employees asked the police for help. According to a court filing, the police responded that dealing with this crowd was “not in their job description,” and they left. Of the two-man security force that Wal-Mart had hired for Blitz Day, only one had shown up, and he was inside the store. Shortly before five, the crowd had grown to about two thousand people. The store’s asset-protection manager, Sal D’Amico, advised Sooknanan not to open the doors, but Sooknanan overruled him. He instructed eight to ten of his largest employees, most of whom worked in the stockroom, to stand at the sides of the vestibule as the outer doors were opened, and be ready to help anyone who tripped or fell.

One of those men was Jdimytai Damour, who lived in Jamaica, Queens; his parents were Haitian immigrants. Damour was thirty-four, and beefy—at six feet five inches tall, he weighed around four hundred and eighty pounds. Friends called him Jdidread, because he wore his hair in dreadlocks. He had been working at Wal-Mart for about a week, as a temporary employee in the stockroom. Like the others in the vestibule, he had no training in security or crowd control. A co-worker had reportedly heard him say earlier, “I don’t want to be here.”

Just before five, the workers realized that a pregnant woman, Leana Lockley, a twenty-eight-year-old part-time college student from South Ozone Park, was being crushed against the glass on the outer doors. The managers slid them open just enough to pull Lockley inside the vestibule. The crowd surged forward, thinking that the store was opening. The workers shut the doors again and braced both sliding doors with their bodies to keep them from caving in, as Sooknanan initiated the festive countdown, a Wal-Mart Blitz Day tradition. Ten, nine, eight . . . At zero, the doors were opened again. There was a loud cracking sound as both sliding doors burst from their frame, and the crowd boiled in.

Dennis Fitch, one of the workers standing at the entrance, was blown backward, through the inner vestibule doors and into the store. Others managed to jump to safety atop the vending machines. Some attempted to form a human chain on the other side of the vestibule, to slow down the crowd rushing into the store. A crush soon developed inside the vestibule, but the people who were still outside, pushing forward, weren’t aware of it. Leana Lockley was carried through the vestibule and into the store by the surge, and she tripped over an older woman, who was on the ground. As she got to her knees, she later said, she saw Damour next to her. “I was screaming that I was pregnant, I am sure he heard that,” she told Newsday . “He was trying to block the people from pushing me down to the ground and trampling me. . . . It was a split second, and we had eye contact as we knew we were going to die.”

Co-workers later testified that Damour was hit by one of the two sliding glass doors. As he went down, the door fell on top of him, and people fell over it. Maybe he got up again to help Lockley, but that’s not clear in camera and cell-phone-video footage of the scene. He just vanishes into the frantic tangle of limbs.

Lockley’s husband, Shawn, was able to pull her out, badly bruised. A healthy girl was born the following April. But though “Big guy down” was broadcast over the walkie-talkies that some of Damour’s co-workers carried, they had to fight their way through the crowd to reach him, and when they got there Damour’s tongue was out and his eyes had rolled back. The cops arrived at 5:05 A.M ., and performed CPR (a cell-phone video made its way to YouTube), without success. Damour was pronounced dead at the Franklin Hospital Medical Center, in Valley Stream, at 6:03 A.M. The coroner’s report did not mention any bruises, fractures, or internal injuries, as it would have if he’d been trampled to death; the cause of death was listed as asphyxia.

Crowds are a condition of urban life. On subways and sidewalks, in elevators and stores, we pass in and out of them in the course of a day, without pausing to consider by what mechanisms our brains guide us through so easily, rarely touching so much as a stranger’s shoulder. Crowds are often viewed as a necessary inconvenience of city living, but there are occasions when we gladly join them, pressing together at raves and rock concerts, at sporting events, victory parades, and big sales. Elias Canetti, in his 1960 book “Crowds and Power,” sees these times of physical communion with strangers as essential to transcending the fear of being touched. “The more fiercely people press together,” he writes, “the more certain they feel that they do not fear each other.” In fact, a crowd is most dangerous when density is greatest. The transition from fraternal smooshing to suffocating pressure—a “crowd crush”—often occurs almost imperceptibly; one doesn’t realize what’s happening until it’s too late to escape. Something interrupts the flow of pedestrians—a blocked exit, say, while an escalator continues to feed people into a closed-off space. Or a storm that causes everyone to start running for shelter at the same time. (In Belarus, in 1999, fifty-two people died when a crowd tried to enter an underground railway station to keep dry.) At a certain point, you feel pressure on all sides of your body, and realize that you can’t raise your arms. You are pulled off your feet, and welded into a block of people. The crowd force squeezes the air out of your lungs, and you struggle to take another breath.

John Fruin, a retired research engineer with the Port Authority of New York and New Jersey, is one of the founders of crowd studies in the U.S. In a 1993 paper, “The Causes and Prevention of Crowd Disasters,” he wrote, “At occupancies of about 7 persons per square meter the crowd becomes almost a fluid mass. Shock waves can be propagated through the mass sufficient to lift people off of their feet and propel them distances of 3 m (10 ft) or more. People may be literally lifted out of their shoes, and have clothing torn off. Intense crowd pressures, exacerbated by anxiety, make it difficult to breathe.” Some people die standing up; others die in the pileup that follows a “crowd collapse,” when someone goes down, and more people fall over him. “Compressional asphyxia” is usually given as the cause of death in these circumstances.

Crowd disasters occur all over the world, and for a variety of reasons. According to a recent paper published in the journal Disaster Medicine and Public Health Preparedness , reports of human stampedes have more than doubled in each of the past two decades. In the developing world, they often occur at religious festivals. In November, hundreds of people died in Cambodia, in a crush that occurred on a bridge in Phnom Penh during the annual water festival; there were reports that the police had fired water cannons at people on the crowded bridge. Thousands have died making pilgrimages to Mecca in the past twenty years, mainly in the ritual called the Stoning of the Devil, which occurs near the Jamarat Bridge; in 2006, three hundred and sixty pilgrims were killed there. In India last month, more than a hundred Hindu worshippers died in a crush in the state of Kerala.

In the developed world, soccer games and rock concerts are the most likely events to generate deadly crowds. In 1989, in Sheffield, England, ninety-five people died after they were caught in a crowd crush at Hillsborough stadium when fans were trying to get into a soccer match between Nottingham Forest and Liverpool. (A ninety-sixth victim was taken off life-support four years later.) At a rock festival in Roskilde, Denmark, in 2000, nine people died after a crowd collapse that occurred near the stage while Pearl Jam was performing in front of an audience of fifty thousand. Last July, twenty-one people were killed at the Love Parade, a free electronic-music festival in Duisburg, Germany, when a crush developed in a disused rail tunnel that led to the festival grounds. With the world’s population increasing, and with more people moving to cities, crowds will become ever larger, and disasters more frequent, unless scientists and safety engineers can figure out how to prevent them from happening.

In the literature on crowd disasters, there is a striking incongruity between the way these events are depicted in the press and how they actually occur. In popular accounts, they are almost invariably described as “panics.” The crowd is portrayed as a single, unified entity, which acts according to “mob psychology”—a set of primitive instincts (fear, followed by flight) that favor self-preservation over the welfare of others, and cause “stampedes” and “tramplings.” But most crowd disasters are caused by “crazes”—people are usually moving toward something they want, rather than away from something they fear, and, if you’re caught up in a crush, you’re just as likely to die on your feet as under the feet of others, squashed by the pressure of bodies smashing into you. (Investigators collecting evidence in the aftermath of crowd disasters have found steel guardrails capable of withstanding a thousand pounds of pressure bent by crowd force.) In disasters not involving fire, panic is rarely the cause of fatalities, and even when fire is involved, such as in the 1977 Beverly Hills Supper Club fire, in Southgate, Kentucky, research has shown that people continue to help one another, even at the cost of their own lives.

So why do we still think in terms of panics and stampedes? In many crowd disasters, particularly those in the West where commercial interests are involved, different stakeholders are potentially responsible, including the organizers of the event, the venue owners and designers, and the public officials and private security firms whose job is to insure crowd safety. In the aftermath of disasters, they all vigorously defend their interests, and rarely are any of them held accountable. But almost no one speaks for the crowd, and the crowd usually takes the blame.

The origins of the term Black Friday are obscure. Some think that it was first used by the police in Philadelphia to describe the snarled traffic and sidewalk hassles that came with the day after Thanksgiving and crowds arriving for the city’s annual Army-Navy game. Others have defined Black Friday as the day that merchants’ balance sheets crossed over into the black. Either way, it is now a de-facto national shopping holiday. On TV, images of people racing through the aisles of stores for sale-priced items, in a sort of American Pamplona, have become as much a part of the day after Thanksgiving as leftovers. Shoppers get discounts, programmers get some lively content for a slow news day, and retailers get free publicity: a good deal for everyone, except for the clerks who have to work that day, breaking up fights among shoppers and cleaning up the mess left behind.

“Im not tracking anything. Im about to ask for your hand in marriage.”

Link copied

There had been injuries on previous Black Fridays, but no one had ever died before Jdimytai Damour went down in the Valley Stream Wal-Mart. His death, and the “Wal-Mart Stampede” that caused it, was the lead story on news channels across the country that evening, and it provoked a vast outcry of horror. In days of commentary that followed, the crowd was widely vilified. The tone of much of the reaction was captured by a letter writer to the New York Post , who blamed “the animals (you know who you are) who stampeded that poor man at Wal-Mart on Black Friday: You are a perfect example of the depraved decadence of society today.”

A Wal-Mart senior vice-president, Hank Mullany, said in a statement, “Our thoughts and prayers go out to the family of the deceased. We are continuing to work closely with local law enforcement, and we are reaching out to those involved.” Investigators would be reviewing video collected from security cameras, and looking at purchases made with credit cards, in an effort to identify individuals who may have witnessed or been involved in Damour’s death. But even if investigators could pick out, amid the flailing limbs and hurtling bodies in the videos, those who had harmed Damour, who could say that he or she wasn’t pushed by the person behind? And, at any rate, the police seemed to be in no mood to “work closely” with Wal-Mart. Rather, they went out of their way to blame Wal-Mart for the incident. Detective Lieutenant Michael Fleming, who was in charge of the investigation into Damour’s death, said at the time, “I’ve heard other people call this an accident, but it is not. Certainly it was a foreseeable act.”

Through the winter and spring of 2009, the Nassau County District Attorney’s office prepared to bring criminal charges against Wal-Mart, for felony reckless endangerment and misdemeanor reckless endangerment. The family of Jdimytai Damour filed a wrongful-death claim against the company. However, in early May, 2009, the county’s District Attorney, Kathleen Rice, announced that her office had worked out a deal with Wal-Mart that allowed it to avoid criminal charges. The company agreed to donate one and a half million dollars to various community projects and to create a four-hundred-thousand-dollar victims’ fund. Wal-Mart also agreed to implement a “crowd management” plan for future post-Thanksgiving Day events at each of its ninety-two New York stores. In return, Wal-Mart would face no charges or criminal liability for the death of Jdimytai Damour. If the company failed to meet the standards set by an independent monitor for three years, the criminal case would be reinstated. Rice, noting that the maximum penalty Wal-Mart would have faced was a ten-thousand-dollar fine, said, “This agreement does more than any criminal prosecution could ever accomplish.” Damour’s father, Ogera Charles, saw things differently. “It’s like if they were driving a car and they hit someone, killed him, and then just walked away,” he told Newsday .

At the end of May, the Occupational Safety and Health Administration cited Wal-Mart for committing a “serious violation” of the General Duty Clause of the OSHA Act. The clause states that an employer must furnish workers with a place of employment that is “free from recognized hazards that are causing or are likely to cause death or serious physical harm to his employees.” In its complaint, OSHA listed the hazards that Damour and his co-workers faced there as “asphyxiation or being struck due to crowd crush, crowd surge or crowd trampling.” The complaint also said that Wal-Mart “did not use appropriate crowd management techniques to safely manage a large crowd of approximately 2000 customers.”

The proposed penalty was seven thousand dollars—not an enormous burden for the world’s biggest retailer, which had total sales of four hundred and five billion dollars in 2010. But Wal-Mart elected to contest the citation, and hired the Washington, D.C., law firm of Gibson, Dunn & Crutcher to handle the litigation. Wal-Mart objected on multiple grounds. First, if crowd crushes and surges were recognized hazards, then why hadn’t a single OSHA General Duty Clause citation ever referred to the dangers posed by crowds before? Wal-Mart also maintained that it had taken steps to protect its workers from the crowd, but it could not have protected workers from this particular crowd. And, finally, the violence caused by the crowd was a police issue and therefore beyond OSHA ’s jurisdiction.

A federal administrative-law judge, Covette Rooney, of the Occupational Safety and Health Review Commission, was assigned to the case, but it did not come to trial for more than a year. Wal-Mart’s lawyers filed twenty pretrial motions and responses, and spent, by OSHA ’s calculations, two million dollars fighting the citation. In all, OSHA lawyers invested around five thousand hours in the case. Why was Wal-Mart fighting a paltry fine so hard? To the extent that the citation could strengthen the Damour family’s civil case, two million dollars could be seen as a worthwhile gamble. Moreover, no retailer welcomed OSHA jurisdiction over how it managed its customers. Casey Chroust, an executive vice-president of the Retail Industry Leaders Association, told me, “The impact of this case is potentially huge. Does it mean I have to hire an event-management staff next time I hold a doorbuster sale? Does this mean every time you have a hot product—a video game, a Harry Potter book, an iPhone—much less a Black Friday sale, you’ll be liable for potential action if you don’t hire crowd management?” Willis Goldsmith, a partner at the New York firm of Jones Day, who has a long history of representing employers on OSHA issues, told me that, along with the problem of defining crowd surges and crushes as recognized hazards, there was the practical matter of defining a crowd. “Ten people could have caused the injuries you saw at Wal-Mart. So is that a crowd?”

OSHA ’s burden was to prove that crowd surge and crowd crush are well-known phenomena, and that crowd-management techniques could have prevented them at the Green Acres Mall. To do that, it needed to find an expert who would testify against Wal-Mart. Most experts in the field consult for private industry—event planners and promoters, venue owners and operators, and, to a lesser extent, large retailers. Even if they agreed with OSHA , testifying against the world’s largest retailer wasn’t likely to be good for business, and many experts wouldn’t do it. But one would: Paul Wertheimer, the sixty-two-year-old self-employed owner of Crowd Management Strategies, who has been called “the marshal of the mosh pit.”

One of the best-documented crowd disasters in the U.S. occurred before a concert by The Who, outside Riverfront Coliseum in Cincinnati, on December 3, 1979. Until then, crowd planning had largely been the purview of fire-safety engineers, who focussed on how to get people out of buildings, in the event of an emergency—not into them. The concert’s promoter, the Electric Factory of Philadelphia, had offered unreserved “festival seating”—people in the front of the line get to be nearest the stage (and, in most cases, no one on the floor has a seat at all, allowing the promoter to sell more tickets but giving the venue far less control over the audience). Hard-core fans began lining up in the early afternoon, and by six o’clock a crowd of eight thousand mostly young people had collected on the plaza outside the entrance, on a bitterly cold night. The band began its sound check at around six-thirty, and played for half an hour. People toward the back of the line, mistakenly believing that the concert was beginning, pushed forward. Some of the people in front pushed back, and shock waves began to ripple through the tightly packed mass. The coliseum staff, thinking that the crowd was attempting to rush the doors and enter without paying, kept most of the doors shut, even after the sound check ended and the opening time had passed.

Later, in a letter sent to the task force assembled to investigate the incident, in which eleven people died, a man in the crowd described what it was like near the doors: “The pounding of the waves was endless. . . . If a wave came and you were being stood upon with your feet pinned to the ground, you would very likely lose your shoes or your balance and fall.” Some people near the doors did go down. “They began to fall, unnoticed by all but those immediately surrounding them. People in the crowd 10 feet back from them didn’t know it was happening. Their cries were impossible to hear above the roar of the crowd. . . . There was a pile of people forming, and all of the people around them were being crushed into the pile, for there was no resistance. If the person in front of you went down, then you would follow for there was no one to lean against.” Then the waves began to carry him toward the pile. “With this realization I began to add to the screaming, ‘They’re going down, they’re going down!’ I yelled repeatedly. . . . A wave swept me to the left and when I regained a stance I felt I was standing on someone. The helplessness and frustration of the moment sent a wave of panic through me. I screamed with all my strength that I was standing on someone. I couldn’t move. I could only scream.”

The media blamed the crowd. The Lexington, Kentucky, Herald-Leader, describing the “surging, primitive mob,” quoted a security guard who said, “Those kids were animals.” Mike Royko wrote a column for the Chicago Sun-Times , entitled, “Cincinnati Barbarism: A Rockwork Orange,” blaming the “barbarians” who “stomped 11 persons to death [after] having numbed their brains on weeds, chemicals, and Southern Comfort.” The promoter, Larry Magid, told Rolling Stone , “After all, we didn’t trample anyone to death, we didn’t step on anyone, and we didn’t push anyone.” Pete Townshend, the band’s leader, said, “It’s rock. It’s not The Who. It’s rock and roll. Everybody—all of us—we’re all bloody responsible.” In the end, no one was held accountable for the deaths.

At the time, Paul Wertheimer was a twenty-nine-year-old public-information officer for the city of Cincinnati. He became chief of staff of the task force that Mayor Kenneth Blackwell appointed to investigate the incident. Wertheimer and some of his staff members spent months travelling around the country, talking to venue operators and promoters and public-safety officials. Among the task force’s recommendations were a ban on festival seating for large indoor events, and a requirement that organizers file a “crowd management” plan, similar to a fire-safety plan, but focussing on ingress as well as egress. The report pointed out that doors and turnstiles in buildings of public assembly were tested only for normal conditions, and failed to take crowded conditions into account. It also called for national standards to better protect crowds. But national standards weren’t created and festival seating wasn’t universally banned. Injuries and fatalities at concerts continued.

As Wertheimer worked at various jobs in event management and public relations, “the Who tragedy kept following me around,” he recalled. “Every now and then, another incident would happen at a concert, someone would get killed, and the reaction was always the same. The industry would say, ‘How could we have predicted this? This has never happened before!’ And of course I would say, ‘That’s not true—it did happen, and here’s a report about it!’ But the industry chose to ignore that. And I thought, Somebody has to step up and do something, because there are ways to prevent these people from dying. And I guess that guy is going to be me. I am going to be the ghost of that Who concert. Those eleven people died so that these lessons could be learned, and I’m going to see they aren’t forgotten.”

Wertheimer began carefully documenting crowd-related incidents in the U.S. and around the world, making the information available to the public. He ventured into potentially dangerous crowds, wherever he could find them, and noted what he saw. In the early nineties, with the popularity of grunge music, mosh pits became common at rock concerts—fans in the front would hurl themselves at one another, and the force would carry them into other fans. Mosh pits are good places to study crowd dynamics, because they reproduce in miniature the shock waves of large-scale crowd disasters. Wertheimer, in his early forties, became a familiar figure at grunge and heavy-metal shows: “the old man in the pit,” in the words of one young fan. “I learned how to stand in the center spot,” he told me proudly, “right in front of the lead singer, three yards from the stage, and to go with the surge, and I developed my ways of getting out of tight spots, which I published in my mosh-survival guide. I worked on my peripheral vision, and learned to recognize when people are in trouble, and to understand what draws them to moshing, and how the band relates to it, and what security does in certain situations—all that stuff.” He established a Web site, Crowd Safe, where he published his reports on crowds, which eventually numbered in the thousands.

“Hes away attending to a personal matter but he did leave a number where he cant be reached.”

As predicted, none of this helped Wertheimer’s career as a crowd-management consultant; his pugnacious personality didn’t help, either. “The industry didn’t want anything to do with me,” Wertheimer told me. In Chicago, where Wertheimer was born, on the South Side, he ran afoul of a concert promoter, Jam Productions, for helping to publicize safety issues at rock concerts. (Wertheimer brought a local news reporter, with a concealed camera, into the mosh pit at a show put on by Jam, and pointed out the unsafe conditions. Jam contends that the footage was misleading.) Jam posted his photograph around Soldier Field, and during a Pearl Jam concert Wertheimer was picked up in a mosh pit by security for apparently shoving a young fan. “Obviously, if I wanted to develop a consulting business, this wasn’t the way to do it,” he told me.

After the deaths of the nine festival-goers during the 2000 Pearl Jam set in Roskilde, Wertheimer was interviewed by a committee set up by the Danish government, and recommendations he made became a part of the committee’s official report, “Rock Festival Safety.” He was delighted when OSHA asked him to testify in the Wal-Mart case. “This is the most important thing I’ve ever been involved with,” he said. “For the first time, you’ve got someone powerful—the U.S. federal government—alleging that this death was preventable, if the crowd had been handled the right way.” Was he anxious about the trial? “I know you can pay a price if you take on a large corporation like Wal-Mart. You have to be willing to suffer the consequences. I don’t have kids to support, or a family; this is the role I take. I’m the only one who would do this. And, hey, I learned to fight on the South Side.”

During the years that Wertheimer was recording his experiences at rock concerts, researchers in academia were trying to figure out models for crowd behavior. In the early nineties, Dirk Helbing, a graduate student in physics at the University of Göttingen, Germany, was looking for a suitable topic for a diploma thesis, when he was inspired by footprints left in the snow after a large event. He saw a pattern in the tracks that suggested the flow of streams, and he came up with a model based on fluid dynamics to simulate crowd movement. By comparing computer-driven simulations with empirical observations of crowd movement, Helbing and his colleagues were able to identify several patterns of collective behavior that emerged from the interactions of individuals in the crowd. These included lanes of uniform walking directions, oscillations of the pedestrian flow at bottlenecks, and “stripes” of intersecting flows. “Such self-organized patterns of motion demonstrate that efficient, ‘intelligent’ collective dynamics can be based on simple, local interactions,” Helbing wrote in a 2010 paper, “Pedestrian, Crowd, and Evacuation Dynamics,” published in the “Encyclopedia of Complexity and Systems Science.”

But Helbing also observed that at certain critical densities, such as occur in a crowd crush, all forms of collective behavior vanish. Shock waves are the result not of collective behavior but of the failure of it. Individuals at the back of a crowd, unable to tell what is happening up ahead, push forward, not realizing that they are injuring the people in the front. Unlike ants and fish and birds, humans haven’t evolved the capability to transmit information about the physical dynamics of the crowd across the entire swarm. Ants, for example, are able to communicate within a swarm using pheromones. Iain Couzin, a behavioral biologist at Princeton University, told me, “With ants, as with human crowds, you see emergent behavior. By using a simple set of local interactions, ants form complex patterns. The difference is that we are selfish individuals, whereas ants are profoundly social creatures. We want to reduce our travel time, even when it is at the expense of others, whereas ants work for the whole colony. In this respect, we are at our most primitive in crowds. We have never evolved a collective intelligence to function in large crowds—we have no way of getting beyond the purely local rules of interaction, as ants can.”

So is there no possibility that a crowd of bodies can be “smart,” in the sense that a crowd of minds can be? Couzin pointed to the role that “leaders” play in the sudden movements of schools of fish, or in migratory herds of animals: only a few of the animals possess the necessary information about where to go, but the others spontaneously follow them. In 2005, he helped design an experiment at Leeds University, led by Jens Krause, in which two hundred people were told to walk randomly around a large hall, while a few people were given specific instructions about what route to take. The researchers found that the “naïve” group followed the informed “leaders,” even though they had no idea, in most cases, that they were following leaders at all. “Leadership does not require verbal communication,” Couzin told me. Studies of disaster evacuations, including the 2001 World Trade Center bombing, have shown that people who follow well-informed leaders might stand a better chance of escape than people who delay or seek their own way out, but in a crowd crush that isn’t going to help much. The leaders will be hemmed in, too.

The Wal-Mart trial took place during six very hot days in July, 2010, in a courtroom in the Jacob Javits Federal Building, in lower Manhattan. Four Wal-Mart employees, who had been at the entrance of the vestibule with Jdimytai Damour, testified. Justin Rice, who had been promoted to department manager before Black Friday, 2008, and who was still working at the store, said that the doors had broken on Blitz Day in 2007, and he had been nicked by broken glass. (Another employee said that the doors came off the hinges in 2005 and 2006, as well.) All the men said that they had never had any training in crowd management before being placed in the vestibule on November 28, 2008, except for “slip, trip, and fall” guidelines—if a customer slips, you help him up—and the “ten-foot rule,” which is if a customer gets within ten feet you are supposed to greet her with “Welcome to Wal-Mart.”

One particularly damning bit of evidence was a video that students from the New York Institute of Technology had chanced to make of a management meeting two days before the Blitz Day event. Rice can be heard raising the matter of the 2007 melee with Steve Sooknanan, the Wal-Mart manager, and saying that people had to be kept away from the doors this year. He says, “Last year was crazy, a lot of people fell, little babies out there and it was cold, I just don’t want that this year.” Sooknanan tells him that this year “we’re going to do it a little differently.” He explains that he had arranged for construction barriers to be placed farther from the entrance and to have additional staff at the door.

Jason Schwartz, the lead trial attorney for Wal-Mart, wasted no time in attacking Paul Wertheimer’s qualifications as a crowd expert—“the dubiously monikered ‘marshal of the mosh pit’ ”:

J.S.: What do you do when you’re in a crowd, Mr. Wertheimer, in order to enhance your expertise? P.W.: I observe the crowd, the crowd dynamics, the crowd behavior, and people in the crowd and talk to people in the crowd to see how they’re feeling, see what’s going on. J.S.: If I did that, would I have the same level of experience in crowds as you do? P.W.: No. J.S.: Why not? P.W.: You’re not an expert in the area of crowd management. J.S.: I see. . . . Your Honor, I would submit that this expert’s qualifications are the same qualifications that everyone standing in this courtroom has.

Judge Rooney responded, “But he has more experience in crowds than I do. I don’t take subways, so I have no idea what it’s like to be in a crowd. Well, I could say, back in my days of college, I took the subway here in New York, and I was very claustrophobic. So I do believe that there is some assistance that, or some value that, is going to be elicited from this case.”

Wertheimer was allowed to continue, and during two days of testimony detailed many crowd measures that Wal-Mart could have taken. He was particularly effective in showing why the construction barriers wouldn’t control the crowd: they were too low to keep people out, and they were flared at the bottom, so that people who got pushed up against the sides fell in.

At the end of six days, Judge Rooney had twelve hundred pages of testimony to deliberate over, which she has done, at a stately pace, for the past six months. Both sides eagerly await the verdict, which is expected soon. If OSHA wins, Wal-Mart will almost certainly appeal—all the way to the U.S. Court of Appeals, if necessary. Still, a decision for OSHA will have enormous symbolic value, because it would be a victory for the crowd.

In the past thirty years, safety officials and designers have learned a lot about crowd management. After the Hillsborough disaster, Britain banned standing terraces in its top two soccer divisions, and introduced “all-seater” stadiums. Some people argued that this changed the atmosphere of the games profoundly, but it also made them safer. An international team of experts, including Keith Still, a professor of crowd dynamics, made recommendations for the redesign of the Jamarat Bridge, in Mecca, and for directing the movement and flow of people. The structure has been altered to provide pilgrims with multiple entrance and exit points, to ease congestion. In Times Square on New Year’s Eve, the police use lightweight metal container pens so that people revel inside a series of small enclaves, rather than as one big mass. Crowd managers use elevated viewing platforms, to see over the crowd, and, if necessary, to communicate with people in the back. Paul Wertheimer has written a booklet, “You and the Festival Crowd,” which has been widely distributed. (Among his recommendations: Keep your elbows akimbo, to protect your chest and give yourself enough breathing room. Don’t fight against the flow of the crowd if you’re trying to get out of it; rather, go with it, and during lulls try to work your way diagonally through the crowd to the perimeter. If you feel faint, grab on to someone, and, if you do fall, try to protect your head.)

And yet, almost anywhere, you can be trapped in a crowd: on a subway platform, at the lighting of the Christmas tree in Rockefeller Center, on the ramps leading down from the upper tiers at Yankee Stadium, in the Halloween Parade in Greenwich Village. One reason last summer’s Love Parade disaster in Germany was so shocking is that it occurred in a country known for efficient crowd management, and yet the early evidence suggests that the organizers and the police made a series of elementary mistakes, including underestimating the number of attendees, using the railway tunnel as both the main entrance to and the main exit from the event, and blocking the flow of concertgoers at pinch points, which allowed the crowd force to build. A full-scale investigation is under way.

A light rain was falling over the parking lot at the Green Acres Mall when I pulled in, at three in the morning, on Black Friday, 2010. The longest line was at Best Buy—it stretched the length of the building and halfway down the other side. The people in front had been waiting for twenty-eight hours. “Wii Bundles,” one man said, when I asked why, as though the answer were obvious. Target also had a long line outside, and there were smaller lines outside Kohl’s and Macy’s. But outside Wal-Mart there was no line at all.

After Black Friday, 2008, Wal-Mart dropped the term Blitz Day, and rebranded its post-Thanksgiving Day sale The Event. In keeping with the terms of its agreement with the Nassau County D.A.’s office, the company employed a crowd-management plan at all its New York stores. In Valley Stream, there were more staff, security, and crowd managers outside the store than there were customers. I snaked through the barricades—metal, chest high, with open bottoms—that had been arranged in a series of tight S curves, passing two viewing platforms, with a man on each holding a bullhorn welcoming me. I entered the vestibule where Damour died, remembering the images of chaos I had seen in the videos of that night. Perhaps the most horrifying aspect of those videos is the sound inside the vestibule: cries of pain, fear, terror, mayhem. But now it was eerily quiet.

This year, like last, the waiting took place inside the store, which remained open all night. Beginning at midnight, the store began distributing tickets for the steeply discounted electronic items, and by three-fifteen they had all been given out. People arriving when I did weren’t happy. “You said the sale starts at five. That’s false advertising,” one irate customer said to a manager. “It’s not me, it’s them,” the manager said, gesturing toward the ceiling. People were lining up anyway for the ordinary sale-priced items, but there was no joy of the hunt in the line. It was just a line. ♦

Recent trends in crowd management using deep learning techniques: a systematic literature review

Open access
Published: 20 June 2024

Cite this article

You have full access to this open access article

Aisha M. Alasmari 1 ,
Norah S. Farooqi 2 , 3 &
Youseef A. Alotaibi 4

909 Accesses

Explore all metrics

Crowd management has become an integral part of urban planning in abnormality in the crowd and predict its future issues. Big data in social media is a rich source for researchers in crowd data analysis. In this systematic literature review (SLR), modern societies. It can organize the flow of the crowd, perform counting, recognize the related works are analyzed, which includes crowd management from both global and local sides (Hajj events—Saudi Arabia) based on deep learning (DL) methods. This survey concerns crowd management research published from 2010 to 2023. It has specified 45 primary studies that accomplish the objectives of the research questions (RQs), namely, investigation of the taxonomies, approaches, and comprehensive studies of crowd management both globally and locally and focusing on the most commonly used techniques of DL. We found both supervised and unsupervised DL techniques have achieved high accuracy, with different strengths and weaknesses for each approach. A lot of these studies discuss aspects of scene analysis of crowds, that are captured by installed cameras in the place. However, there is a dilemma regarding exploiting data provided on social media to use in the crowd analysis domain. Which we believe that the analysis of big data may raise crowd management to the upper level of enhancement. To this end, motivated by the findings of this SLR. The primary purpose of this review is strived to illustrate obstacles and dilemmas in crowd analysis fields to provide a road map for future researchers. Furthermore, it aims to find research gaps existing to focus on it in the future studies. The results indicate that the lack of Hajj research, especially in sentiment analysis and the study of the pilgrims' behavior.

A Review of Deep Learning Techniques for Crowd Behavior Analysis

Video analytics using deep learning for crowd analysis: a review

Convolutional neural networks for crowd behaviour analysis: a survey, explore related subjects.

Artificial Intelligence

Avoid common mistakes on your manuscript.

1 Introduction

A crowd is defined as a gathering of humans in the same area. If the number of individuals exceeds normal conditions, congestion becomes a concern regarding safety, health, or what may affect human choices due to herd culture. The concept of congestion differs depending on the culture of communities.

For example, gathering over 100 people in India is considered normal, while in other countries like Canada it may be considered a crowd [ 1 ]. Analysis, detection, management, and monitoring of crowds are a growing trend in computer sciences to study the behavior of human crowds in any event. Human gatherings may happen due to religious rituals (such as Hajj for Muslims at Makkah, Saudi Arabia, and Kumbh Mela for Hindus at Haridwar, India), sporting events (such as FIFA World Cup and Olympic Games), concerts, or annual carnivals (such as Carnival Parade and Riyadh Season [ 2 ]). Moreover, demonstrations, popular protests, and political or social riots (such as a political rally in Los Angeles [ 3 ]) are considered gatherings that may impact a crowd risk index and bring many unexpected reactions. Each type of human gathering has its own features: purposes of the individuals, behavior, place, and time. To avoid accidents, the organizers must perform prior analyses and studies for the mass gatherings.

Crowd analysis is one of the most crucial tools for crowd management [ 4 ]. Hence, crowd management requires profound and comprehensive plans in advance and flexible and agile strategies to take vigorous and accurate action if unexpected incidents occur.

Examples of common general issues in crowd management in the context of urban planning for smart cities are traffic, crowded pedestrians, pollution, energy consumption, etc. In urban planning for smart cities, it can be possible to predict crowd behavior abnormalities and the future evolution of these situations in order to prevent them and do the best decision-making and planning. In addition, the capacity to monitor, control, and predict the behavior of crowds is a fundamental enabling driver. Where predicting such notifications can be of effective help in a large variety of situations, such as organizing events, organizing pedestrians, managing situations of emergency, or even tracking how the pandemic spread through the urban areas. A violation of crowd management will lead to several consequences that may result in a loss of lives or property, besides the loss of people's confidence in the organization responsible for organizing the event in the future.

Billions of people around the world now have accounts on social media platforms to freely express their beliefs, opinions, or impressions about some things. This huge streamed data gives an opportunity for researchers in the data analysis domain to explore about behavior of people through their text content [ 5 ]. We believe big data may open other horizons in crowd management. Abnormal behavior detection or crowd-counting is now possible through these data. However, a lot of works of crowd management lack to attention the textual data analysis aspect of social media, especially in local crowd management works.

This review has conducted extensive examinations in this area, however, a lot of works for crowd management still have limited in using one particular data namely visual data. The main drawbacks and limitations faced by current crowd management are discussed as follows:

Existing works for crowd management are currently recurrent, meaning they do not take into account the collected data sources changes about the people during the behavior detection. Hence, current models may not fully apply on the multi-dataset, limiting their effectiveness with similar scenarios.

Collecting the datasets for existing models requires consideration of equipment or hardware such as installed cameras, live streaming channels, etc., that increases the cost of running the processes and maintenance besides the computational cost of the models.

Most of the existing research uses the same video dataset to study the behavior detection of crowds. Therefore, limiting their model's effectiveness with learning new patterns.

There is no dependency on data of social networks as one of the sources of data collection.

For these reasons, the objectives of this SLR are an overall survey of the concept of crowd management from two perspectives, crowd management in various world events and Hajj events especially. The authors endeavor through this work to make it a great reference point for other researchers in the crowd management domain. This systematic review provides a theoretical understanding of deep-learning techniques used in the various branches of crowd management. The review will also highlight factors that impact crowd modeling works such as limited patterns of datasets, applications generalizability, and evaluation metrics. Moreover, the review highlights the importance of urban planning integration, which leads to improving the quality of life for individuals and society. This work emphasizes is also necessary to draw attention to exploiting the various big data in social media as an important tributary of building novel datasets with a diversity of knowledge and patterns. The main contribution of the paper is to examine and summarize the state-of-art technologies and methodologies in the behavior detection of a crowd to apply them in Hajj research to improve the experience of pilgrims and provided services. Hence, the review scope determines three main points as follows: First, our survey summarizes the various technologies, approaches, and models that have been utilized to design and execute solutions to detect behaviors that control and monitor the crowds during any world events. Next, the survey summarizes and review the methods used for Hajj issues. Last, our survey covers shortcomings or defects in previous Hajj studies and how other research related to various crowd management may help improve the management, monitoring, and control of crowds during the Hajj season.

This paper aims to highlight these three points through a comprehensive literature survey and focuses on crowd behavior detection for crowd surveillance and prediction. The purpose of RQs is to give a high level of precise topics which is extremely focused on examining the previous literature. The RQs have extreme significance in an SLR, due to controlling the distinguishing and identification of primary research needed to be involved in the review. Consequently, well-defined, logical, interesting, and relevant research questions should be articulated [ 6 ], at the discretion of the authors [ 7 ]. The review's contributions will answer the following research questions (RQs):

RQ1: What is the taxonomy of crowd analysis for Deep Learning-based works?

RQ1 aims to find the taxonomies of previous studies based on deep learning (DL) approaches. The answer is explained in Sect. 3 : “Related Work.”

RQ2: What are the approaches used in crowd management works?

RQ2 aims to identify the DL approaches used for crowd management at various places around the world. The answer is explained in Sect. 4 : “A Comprehensive Study of Crowd Management.”

RQ3: What are the approaches used in Hajj crowd management works?

RQ3 aims to identify the DL approaches used for crowd management during the Hajj season. The answer appears in Sect. 4.6 : “Crowd Management at Hajj Event.”

RQ4: What are the most commonly used techniques and algorithms in prior works? and the challenges faced them?

RQ4 aims to discover the most used techniques and algorithms of DL in prior works for crowd management at global around the world and local scope during the Hajj season. The answer is explained in Sect. 5 : “Analysis of Comprehensive Study of Crowd Management”.

In conclusion, a systematic review of the research studies in terms of global perspectives on crowd management can help provide insights into the scope and development of this field in Hajj events and establish a comprehensive conceptual framework, which can ultimately improve the pilgrims' experience and the religious rites practices comfortably. The rest of this paper is organized as follows. Section 2 explains the research methodology. Taxonomies of previous studies based on DL approaches appear in Sect. 3 . A summarization of the current methodologies of crowd management at various events appears in Sect. 4 . Section 5 summarizes the current approaches utilized during the Hajj events from 2010 until 2023. Section 5 displays analysis of comprehensive study of crowd management also analysis of crowd management at hajj event. The authors discuss gaps and directions in Hajj studies and compare them with state-of-the-art research on other events in Sect. 6 . Finally, Sect. 7 presents a conclusion of the survey and discuss future work approaches are presented.

2 Research methodology

This section offers the methodology followed to complete this research. The paper [ 8 ] has presented several steps for writing a systematic literature review (SLR). An SLR draws a thoughtful methodology to determine the mechanisms of exclusion and inclusion criteria for scientific papers articles. Moreover, the guidelines of SLR identify gaps in current research and extracts final results based on our RQs. This review was performed in four phases, and the following sections explain each phase. Figure 1 shows the review protocol that illustrates the plan to complete this paper.

Protocol of review to complete this paper

2.1 Phase (1): preliminary search

Initially, verification of previous related work. The authors checked that no SLR covers the topic of crowd management by the analysis of textual data of users. The enormous amount of data spread every second across various Social Media Platforms (SMP), such as Twitter, Facebook, Instagram, and many others, is adequate evidence that aspect extraction of textual data to study it has become urgent. Therefore, the authors use this review of the outputs from past related works to address their gaps and shortcomings.

Second, identification of relevant online databases. The authors selected the superior databases that are interested in computer science: SpringerLink, Science-Direct, IEEE Xplore Digital Library, ACM Digital Library, MDPI, Google Scholar and Web of Sciences. Next, the authors determined the starting and ending publishing dates for the articles in the review. This review selected 2010 as the starting date and 2023 as the ending date. This timeframe was chosen because it is the growth of AI research. In the early 2010s, researchers began to use neural networks for speech recognition and image processing, which has significantly improved performance and then spread neural networks widely in the commercial, healthcare, finance, transportation, and crowd control fields. In 2013, the field of computer vision began to transition using neural networks. The same transition occurred in natural language processing in 2016 until today [ 9 ]. In the future, similar revolutions will occur in visual robotics and many other AI fields. The searches were narrowed to journals published during the desired span. Table 1 presents the number of scientific paper articles obtained from each database and clarification for the initial and final results of the search.

Third, detection of keywords and their synonyms used in crowd management. Keywords of research that have been applied for finding articles in these databases are as follows: Crowd AND (“Management” OR “Analysis” OR “Tracking” OR “Monitoring” OR “Controlling” OR “Counting” OR “Density Estimation” OR “Abnormality Detection” OR “Behavior Analysis” OR “Crowd flow” OR “Mass Gathering” OR “Congestion Analysis Detection” OR “Predicting Human Behaviors” OR “Pedestrian”) AND (“Sentiment Analysis” OR “Opinion mining”) AND (“Deep Learning” OR “Machine Learning” OR “Convolutional Neural Network” OR “CNN”) AND (“Social Media” OR “Twitter”) AND (“Hajj” OR “Makkah” OR “Mecca”).

2.2 Phase (2): investigative search

Criteria of inclusion and exclusion. We selected strict criteria to pick studies to be included in our review and those that must be excluded. The objective of inclusion criteria is to choose all papers describing the concept of opinions mining of crowds through DL techniques. Otherwise, it will be exclusion criteria of papers in order to limit the scope of the review and remain focused on the targeted RQs. Inclusion criteria are as follows:

Papers that were published from the year 2010 to 2023.

Papers written in the English language.

Papers selected for publication in a journal.

In terms of exclusion criteria are as follows:

Papers that are from a conference or a book.

Papers that do not extract specific databases.

Duplicate papers.

Papers that contain irrelevant keywords.

Figure 2 illustrates the criteria of exclusion and inclusion followed for this review.

Criteria of exclusion and inclusion for this review

2.3 Phase (3): study quality assessment

Quality assessment (QA) of selected studies is a critical strategy for data synthesis and analysis to avoid bias and increase the selection of literature. The QA questions estimate the relevance, truthfulness, and rigorousness of the selected studies. Every one of the questions has only three optional answers derived from the study in [ 10 ], where “YES” = 1, “NO” = 0, and “Partly” = 0.5. as shown in Table 2 . Besides the QA questions, it has placed other criteria to prevent potential biases. For instance, clarification of studies included and excluded accurately. Comprehensive examination during the selection and publication stages several times. Formulating review protocols according to the sober methodology [ 8 ]. The assessment selection was from one of the researchers of this paper. The researchers have followed the mentioned standards rigorously to avoid the dominance of individual personal opinions and potentially biased decisions. The included papers should be achieved at least 2 of QA, otherwise, it will be overridden, as shown in Table 3 .

2.4 Phase (4): analysis of search

Transparency during the assessment process is conceived as a non-functional quality of the stakeholders of projects. Transparency is an essential factor that can be performed to ensure the stakeholder's satisfaction with the quality of assessment [ 11 ]. Therefore, transparency requirements should be clarified regarding the inclusion and exclusion criteria that are used for selecting the primary studies, which have to fulfill them to sustainability for quality and transparency of research. Consequently, our methodology identified 45 studies that we applied to the assessment of objectives of this paper. Figure 3 illustrates the peak appearance of research in the publication year 2021. Whereas Fig. 4 displays the percentage of papers obtained from each database.

Distribution of the papers from 2010 to 2023

Percentage of papers obtained from each database

According to our criteria of exclusion and inclusion papers, as shown in the Figs. 3 and 4 , the papers started spread from 2014 to 2023, the researchers note that the publication was at the highest levels in 2019, 2020, and 2021 years. In addition, the SpringerLink database was achieved highest published, whereas ACM database was got the lowest published than other databases. Finally, after applying the above filters of standards did not obtain unique papers in both Google Scholar and Web of Science. Most of the existing papers were duplicates of papers in another publishing database or did not meet our requirements and standards.

3 Related work

Since the last decade, the preceding reviews have illustrated that crowd analysis is studied from several different aspects. For instance, there are computer science [ 12 ], sociology-based [ 13 ], biology-based [ 14 ], and physics-based [ 15 ] approaches. Some of these works concentrate on the research axis, and others concentrate on various sides of the research axes as subtopics. In terms of computer science, there are two main types: traditional approaches from the period of pre-DL methods and DL methods [ 16 ]. DL techniques are a valuable addition to constructing the ideal models in many fields like Defect Prediction in Software (DeP) [ 17 ], improving Search-Based Software Testing (SBST) [ 18 ], improving the mechanisms of Detection of DDoS Attack[ 19 ], remotely imagery classification for unmanned aerial vehicles (UAV) [ 20 ]. Generally, achieving high-level intelligence, high robustness, high accuracy, big data, and low power consumption for artificial intelligence approaches are considered the critical challenges that faces the researchers. The authors in [ 21 , 22 , 23 , 24 ] have sought to address these issues. In this section, our review discusses the most other important reviews. Those that focus on the DL side and large datasets. DL algorithms are more properly suited and effective to address concerns related to the variety, volume, and accuracy of big data analytics. Furthermore, DL algorithms inherently exploit the availability of enormous amounts of data to explore and understand the higher-level complexities of various data patterns. Thus, minimizing the need for human experts to extract features from data [ 25 ].

The reviews aim to offer a panoramic vision of crowd analysis in the deep learning domain. Each previous survey was studied and organized into subsections to classify its authors.

Grant and Flynn [ 26 ] divided crowd analysis into two wide classes, crowd behavior analysis and crowd counting, which include several subsections. Crowd behavior analysis has four subsections: abnormal behavior analysis, dominant motion extraction, crowd analysis and tracking, and group behavior analysis. It focuses on behavior detection of individual scenes at first. Then, it describes group behavior within a crowd, crowd motion, and detection of an abnormal event. On the other hand, crowd counting contains six subsections: density mapping, joint detection and counting, line counting, texture-level analysis, object-level analysis, and pixel-level analysis. It focuses on behavior detection of individual scenes at first. Then, it describes group behavior within a crowd, crowd motion, and detection of an abnormal event. On the other hand, crowd counting contains six subsections: density mapping, joint detection and counting, line counting, texture-level analysis, object-level analysis, and pixel-level analysis.

It discussed the metrics used to estimate the density of a crowd, the Level of Service (LoS), and traffic flow. Moreover, they displayed datasets available according to crowd activity video research. Datasets fell into five categories: crowd counting (UCF_CC_50 dataset [ 69 ], UCSD dataset [ 70 ], and WorldExpo’10 Dataset [ 71 ]), group detection (Collective Motion dataset, The Museum Visitors dataset, student003 dataset, The Mall dataset [ 72 ], and the Grand Central Station dataset), behavior understanding (PETS2009 dataset [ 73 ], Collective Activity dataset, and The Unusual Crowd Activity dataset), holistic crowd movement (Chinese University of Hong Kong dataset (CUHK) [ 74 , 75 ], The Meta-Tracking dataset, Data-Driven Crowd Analysis dataset, and Crowd Segmentation dataset), and synthetic (The Agoraset dataset, Seven Environments/scenes).

Tripathi et al. [ 1 ] concentrated on studies that included Convolutional Neural Networks (CNNs). The authors have divided the previous studies into four classes: The first class summarizes influential portions of the CNN for handling crowd behavior analysis. The second class summarizes the primary studies proposed that focus on CNNs. The third summarizes studies that use CNNs incorporated with other architectures from deep learning. It includes four types, crowd counting, crowd density estimation, crowded abnormality analysis, and crowded scene analysis. The fourth summarizes studies that use CNNs to extract features and classifiers. Moreover, the authors highlighted opportunities, features, and challenges for future research in the crowd analysis domain. Furthermore, the authors displayed some of the datasets used in CNN-based crowd analysis: WorldEx po10, PETS2009 [ 73 ], WorldExpo’10 [ 71 ], Pedestrian dataset, UCLA, Dyntex++, DynTex, (WWW) crowd dataset, BEHAVE, NUS-HGA, UCF_CC_50 [ 69 ], ShanghaiTech [ 76 ], UMN [ 77 ], Mall [ 72 ], Rare Events Dataset (RED) [ 78 ], and City Dataset [ 79 ].

Li et al. [ 80 ] summarized the main concepts of crowd behavior analysis in terms of the Crowd Dynamics concept. It considers a crowd as either a set of individuals such as the Social Force Model or a fluid such as concepts of thermodynamics and statistical mechanics by computer vision. The survey divided the reviewed studies into three classes, anomaly detection, motion pattern segmentation, and behavior recognition. First, crowd motion pattern segmentation analyzes motion patterns in areas of crowded scenes. Several methods have been proposed based on the cluster of the motions or segment principle. For instance, flow-based segmentation, similarity-based clustering, and probability-model-based clustering. Next, crowded anomaly detection has been classified into two sections, global anomaly detection and local anomaly detection, i.e., where does the anomaly occur? Does the scene include an anomaly case or not? Lastly, crowd behavior recognition is classified into object and holistic-based.

Kiran et al. [ 81 ] discussed the detection and prediction of anomalies by defining rare events and detecting unseen objects. Furthermore, the authors present the related works that used DL, unsupervised and semi-supervised methods for anomaly detection in video scenes. They classified their survey according to detection criteria and types of models (deep generative models, predictive models, and reconstruction learning models). Each of these types has several subtypes. Representation learning for reconstruction uses models and methods of normal behavior in surveillance videos to represent deviations in poorly reconstructed anomalies. Examples include principal component analysis, autoencoders, convolutional autoencoders (CAEs), CAEs for video anomaly detection, contractive autoencoders, and other deep models (like stacked DAEs (SDAEs), de-noising autoencoders (DAE), and deep belief networks (DBNs)). Predictive modeling contains four subsections, composite model, convolutional Long Short-Term Memory (LSTM), 3D-autoencoder and predictor, and slow feature analysis (SFA). SFA is used to view video frames as time series or temporal patterns to predict the existing frame or its encoded representation utilizing the previous frames. Lastly, Deep generative models consist of eight subsections: Generative vs. Discriminative, Variational Autoencoders (VAEs), Anomaly Detection Using VAE, Generative Adversarial Networks (GANs), GANs for Anomaly Detection in Images, Adversarial Discriminators Using Cross-Channel Prediction, Adversarial Autoencoders (AAEs), and Controlling Reconstruction for Anomaly Detection. They are employed to model the probability of samples of normal video in a deep learning framework.

Bendali-Braham et al. [ 16 ] proposed a novel taxonomy for crowd analysis that includes two branches, crowd behavior analysis and crowd statistics. Crowd statistics determine the number of people currently in a scene. It includes two subbranches, crowd counting and density estimation. Crowded scene analysis is divided into crowd behavior recognition, motion tracking and prediction, and group behavior recognition for human behavior analysis in a crowded scene. Furthermore, crowd activities and motion patterns are described in video scenes and when crowd statistics determine the LoS. Al-Shaery et al. [ 82 ] tackled an inclusive review of crowd management, from the discovery of crowded places to crowd monitoring and management. They focus attention on systems of crowd management that require a well-designed decision support system (DSS), as well as the systems that have early warning capabilities to realize the primary goal of gatherings which is crowd safety. They divided their taxonomy into two branches: crowd detection and crowd monitoring and tracking analysis. The last section includes the crowd management and control stage that leads to the crowd DSS stage. They considered the crowd management stage as the intermediate between monitoring and the Crowd DSS stage. Ebrahimpour et al. [ 62 ] reviewed the studies of crowd analysis based on various data sources. They divided their taxonomy into three classes, crowd social media analysis, crowd spatiotemporal analysis, and crowd video analysis with some subsections. Crowd spatiotemporal analysis uses a data source generated by transportation that is monitored with Global Positioning System (GPS), such as shared bikes or buses. In terms of crowd social media analysis, it exploits check-in data that have been taken from geo-tagged social microblogs for crowd analysis. The data analysis process contains four steps, discovery, gathering, preparation, and analysis. Finally, crowd video analysis includes two sections with subsections inside them: crowd video behavior analysis (microscopic modeling and macroscopic modeling) for generating trustworthy trajectories for pedestrians as well as crowd video action recognition (single person action recognition and group activity recognition) for single or group activity surveillance, tracking people, objects, sports video analysis, and action recognition.

In summary, the studies and surveys above used various taxonomy according to their perspectives. Most of these studies focused on crowd behavior and motion analysis based on the captured video scenes. One of them focused on crowd spatiotemporal analysis based on GPS data, owing to the ability to collect data automatically remotely by mobile sensing and mobile computing [ 83 ]. Obviously, there is no exploit on social media data, this paper investigates this scope with the best technologies. Our review is distinguished from others that we study general cases of crowd management and analysis, in addition to local studies in the Hajj season to discover the flaws and difficulties facing the Hajj authorities in order to avoid disasters and accidents among crowds.

4 A comprehensive study of crowd management

Literature that discusses crowd management in various past universal events. Through a methodical literature review, they have classified crowd analysis into several types according to the purpose of the study. Crowd scene analysis, social media-based analysis, and crowd sound emotion recognition are the main types of crowd management. Each one has some subsections below, according to our taxonomy in Fig. 5 . Studies discussing the analysis of crowds with various purposes employ DL algorithms. Researchers seek to use the newest of these technologies to achieve the highest performance and accuracy possible. Table 4 illustrates the statistics for papers obtained from each subsection.

Taxonomy of crowd management

4.1 Crowd detection

To avoid accidents, it is crucial to know when the people will gather. Then, the organizers must perform in-depth prior analyses and develop comprehensive plans for these mass gatherings. Crowd analysis is a vital tool for crowd management [ 4 ]. Ordinarily, there will be an advance notice for well-known human gatherings, either religious, sports, carnival events, or always-crowded places such as airports, train stations, stadiums, etc.

Every human gathering has special features regarding the purpose, location, and time as well as the behavior of the people, their beliefs, affiliations, and race. For instance, in 1987, a group of Iranian pilgrims rioted during the performance of the Hajj rituals at Makkah, Saudi Arabia, and, as a result, 402 people were killed and injured [ 61 ]. To give another example of religious events in India, Hindus gather to bathe at the Ganga River, Saraswati River, Kshipra River, and Godavari River, where heavy crowds are expected at specific times. On 31 December 2014, on Shanghai New Year’s Eve, there was a stampede, resulting in 36 individuals killed and 47 others injured [ 60 ]. Table 5 summarizes the tragedies that happened previously. Therefore, it is critical to adopt crowd management and propose rigorous and flexible strategies to prepare for unforeseen occurrences at any time. If crowd management fails, it will lead to a loss of lives or properties.

4.2 Crowd statistics

Crowd counting and density estimation are characteristic types of crowd analysis. Calculation of crowd counting, and density can be beneficial in planning crowd security and safety. If the crowd size can be estimated at crowded places, such as temples, stadiums, airports, or metro stations, in advance, it would be extremely beneficial for planning alternate strategies for crowd control.

4.2.1 Crowd counting

Several methods have been developed for crowd counting, which include three classes under the methodologies of DL: (1) CNN-based methods [ 36 , 42 ]; (2) detection-based methods [ 87 ]; and (3) regression-based methods [ 88 , 89 ]. Briefly, detection-based methods utilize detection algorithms, which consider that a crowd consists of the sliding-window detector and individual entities to compute the number of object instances in the detected image [ 33 , 87 ].

Regression-based methods exist to solve the problem of occlusion. The main ideas of this method are learning a density map and extracting its features from an image to estimate crowd density [ 33 , 88 , 89 ].

Lastly, many works have been developed by CNN-based methods in the crowd counting field due to their successful applications in computer vision.

Kang et al. [ 27 ] proposed an adaptive convolutional neural network (ACNN)-based model for counting. It improves the counting precision compared to an ordinary CNN with a similar number of parameters.

Marsden et al. [ 28 ] had developed a previous model in [ 76 ] of convolutional crowd counting for the high-density crowd. They added several contributions, including a training set increase to minimize redundancy between samples of training to improve counting performance. They also use a single column, deep, fully convolutional network (FCN) for analyzing images with any aspect and resolution ratio.

Sheng et al. [ 54 ] proposed a framework based on locality-aware features (LAF) integrated with CNN features to capture more semantic spatial and attributes of the image. Furthermore, they used a vector of locally aggregated descriptors (VLAD) which consider the weights of the coefficients.

Hu et al. [ 47 ] used a convolutional neural network (convNet or CNN) structure to extract features of a crowd in a single image to estimate the crowd count. Their approach was based on CNN and appropriate for a mid-level or high-level crowd. Similarly, Kumagai [ 29 ] adopted CNNs with fixed weights to reduce the fault rate when counting a crowd.

Dai et al. [ 84 ] proposed improved approaches to crowd flow prediction, whose goal is to count the incoming and outgoing numbers of people in urban regions. The approaches were based on a spatiotemporal attention mechanism with a simplified deep spatiotemporal residual network. The first one captures information about the spatial correlations on crowd flows and finds the regions with positive impacts. The second one reduces training time and gives the best prediction performance compared with similar approaches.

Gong et al. [ 30 ] used existing images on social media to estimate the number of people in crowds at city events. This study is the first to count crowds from this side, unlike prior studies that used datasets from popular sources such as video surveillance data. They constructed a novel dataset of images collected from social media for diverse events and major activities in the city. Each image is annotated with its characteristics and the size of the crowd. They applied four methods of two types, direct methods (Faceplusplus and Darknet Yolo) and indirect methods (Cascaded method A and B), to crowd size estimation analysis. The results showed that direct methods achieve higher accuracy than indirect methods. Specifically, Darknet Yolo achieves the best accuracy in estimating the crowd size level (72.01%) and the number of people (38.09%). This study provides a novel method to count people via the advantage of their visual posts on social media.

Huang et al. [ 31 ] solved the problem of noise in the areas with different densities, which appeared in a previous study that used a multi-column convolutional neural network (MCNN) method. The authors proposed a novel method named a segmentation-aware prior network (SAPNet). Using a map of coarse head-segmentation, they produced a map of high-quality density without noise. SAPNet contains two networks, CR-CNN as the back end and FS-CNN as the front end. They are a crowd-regression convolutional neural network and a foreground-segmentation convolutional neural network, respectively. FS-CNN produces a map of coarse head-segmentation, then this map is inputted to CR-CNN to perform a highly accurate crowd counting to produce a high-quality density map. The four datasets that tested their approach were WorldExpo’10 [ 71 ], UCF-CC-50 [ 90 ], UCSD [ 70 ], and ShanghaiTech [ 76 ]. It has achieved high performances on the UCF-CC-50 and ShanghaiTech part B datasets. However, the WorldExpo’10 dataset [ 71 ] was unsuitable for their method because the raw images are of low precision. Furthermore, a poor Canny-edge map can lead to the generation of a faulty segmentation map. This study succeeds in an efficient solution to the problem of noise in areas with different densities. It will be very beneficial in high-congestion places such as train stations, stadiums, religious gathering.

In the same context, Jiang et al. [ 32 ] produced a novel PSDENet method, the people segmentation-based density estimation network. At first, the PSDENet model performs learning and pre-training on virtual synthetic data, then, it transfers these tests to real data. The proposed method has proven effective even though it uses two independent networks, PSDENet and people segmentation network (PSNet). It requires the consumption of much computation.

Zhang et al. [ 33 ] proposed a two-task convolutional neural network (T 2 CNN). It is a novel method for crowd counting that concomitantly learns two tasks, the density map estimation of images and the classification of the tasks of dense degree. Each image has different degrees of density, and local regions inside them have different degrees of density. Determining the density degrees of images helps the estimation of the density maps. For this purpose, researchers incorporate the module of T 2 CNN with dense degree classification (DDC). T 2 CNN takes the scale of the adaptive CNN as the density maps estimator, then classifies images into several categories based on degrees of density. Therefore, that model is an efficient way to treat the perspective and scale variations in crowd images, according to experimental results performed on common datasets: WorldExpo’10 [ 71 ], UCF_CC_50 [ 69 ], and ShanghaiTech [ 76 ].

Shang et al. [ 34 ] developed a new architecture to deal with the perspective variation problems for estimating the number of people in images on the web. The proposed approach has two-stage processing: policy network and count network. A policy network is an estimation of perspective by a regular CNN, while a counting network is a normalization of perspective for the input patches into a scale-specific CNN. Then, given the arranged inputs, they adjusted the scale-specific counting network and their approach to deal with a large perspective variation in web images. In this context, the evaluation metrics were used to verify the model of Xu et al. [ 91 ], which gives an average enhancement of 4.68% of Grid Average Mean Absolute Error (GAME), 6.7% of Mean Squared Error (MSE), and 3.68% of Mean Absolute Error (MAE). Also, their experiments were performed on datasets following UCF_CC_50 [ 69 ], UCF-QNRF [ 92 ], RGBT-CC [ 93 ], and ShanghaiTech [ 76 ].

Jiang and Jin [ 35 ] discussed estimating high-quality crowd density maps and counting crowds by revisiting the design of CNNs to get high-quality density maps as well as high resolution on datasets of crowd counting. For instance, these datasets include UCF_CC_50 [ 69 ], UCSD [ 70 ], and ShanghaiTech datasets [ 76 ]. Their proposed method, multilayer perception counting (MPC), realized high results in a high-quality density map, which is better than counting the crowd. Their method relies on diverse deep supervision (DDS) rather than general supervision, which uses all the intermediate layers or hierarchical in the network. Moreover, MPC is considered the ideal way for cases requiring prediction in real-time.

Khan and Basalamah [ 36 ] proposed a unified model to detect human heads in visual images for crowds using regression models with CNNs. The model is based on DenseNet, which contains 174 layers. It handles a wide range of scale differences by integrating scale-specific detectors within the network. Therefore, the network parameters are improved in an end-to-end fashion. The model was applied to difficult benchmark datasets, such as UCSD [ 70 ] and UCF-QNRF, and achieved the best results.

Liu et al. [ 52 ] proposed a global density feature to add to the multi-column convolution neural network (MCNN) to improve its performance using the cascaded learning method. This model differs from existing works because it concentrates on uneven crowd distribution. Furthermore, deconvolutional layers and the max pooling were utilized to generate a thorough density map and to restore the missing details of the accuracy of the density map during the down-sampling process. The results of experiments prove that this model has higher accuracy and stability when applied to ShanghaiTech [ 76 ] and UCF_CC_50 datasets [ 69 ].

Kizrak and Bolat [ 4 ] used video images or static images to estimate the number of people in a crowd by utilizing CNN with modules of capsule network-based attention. They have proposed a 75,442 VOLUME to crowd analysis using a CNN and two-column cascade and CapsNet as an attention module. The positive impact of the Capsule attention was proven to detect the number of people in images of a crowd. However, this method is still not effective in terms of computational complexity.

Elharrouss et al. [ 53 ] provided two contributions, a new method using CNN and the creation of a novel crowd counting dataset taken from the Football Supporters Crowd (FSC-Set). It contains 6000 annotated images of various scenes. FSC-Set can be used for other domains such as localization of individuals, image supporter recognition, and face recognition. The proposed method named FSCNet used several modules: channel-wise attention, spatial-wise attention, and context-aware attention for crowd counting. The results were satisfactory on all the datasets. This research provides a solution to counting people in crowded places based on several attributes. This method can be contributed to aid other studies of crowd counting.

Khan et al. [ 55 ] developed a framework using end-to-end semantic scene segmentation (SSS) based on CNN for counting people in a densely crowded image. The framework consists of three components: Density Estimation (DE), classification using optimized CNN, and SSS. Moreover, to solve the problem of scaling variations in images, they used four fields that had sixteen filters to feed output at every stage. Their method has validated four standard datasets such as Shanghai Tech, World Expo, NWPU_Crowd [ 94 ], and UCF_CC_50 [ 69 ]. Furthermore, they claimed that the crowd counting domain is still an immature research area due to limited data in deep learning.

Zou et al. [ 67 ] proposed a model to address ignoring the massive temporal information among consecutive frames when process each video frame independently. The model namely, temporal channel-aware (TCA), it realizes exploiting the temporal interdependencies between video sequences through fusion of 3D kernels of convolution in order to encode local spatio-temporal attributes.

Du et al. [ 68 ] redesigned a classical multi-scale neural network to treat challenging of crowd counting. The scheme merges multi-scale density maps. The network uses both the local counting map and the crowd density map to optimization. The experiments results proved that the novel scheme fulfills the state-of-the-art performance on five public datasets such as UCF_CC_50, JHU-CROWD++, ShanghaiTech, Trancos, and NWPU-Crowd.

At the end, Most of studies above developed their architectures based on CNN features to count crowd. Moreover, they used the famous benchmarks datasets such as Shanghai Tech, World Expo and others to perform experimentation on these architectures.

4.2.2 Crowd density estimation

Density estimation of a crowd is an extended part of crowd counting. Density computation is important to support preset plans and strategies to avoid overcrowding. The authors of [ 51 ] discussed the flow patterns of a crowd. They used an unsupervised methodology to cluster people patterns in large public infrastructures. The proposed approach has been applied to an international airport. Their approach successfully summarized the representative patterns and provided the required data for airport management.

The work of [ 44 ] proposed a model to estimate crowd density via an optimized ConvNet. The model has two ConvNet classifiers to improve its speed and accuracy. In the same context, the work of [ 63 ] used LSTM-combined Node2Vec graph embedding to extract spatial features.

4.3 Crowd scene analysis

Crowd Scene Analysis is most important to study normal or abnormal human behavior. This aspect includes Crowd Motion Analysis and Tracking using the most common approach is video surveillance to detect alarms and anomalies.

4.3.1 Crowd monitoring and tracking

[ 95 ] developed a new framework for an online gating neural network. It consists of two phases: the offline training phase and the online predicting phase. In the first phase, their training set is trained daily using a gated recurrent unit-based predictor of human mobility. In the second phase, they constructed an online adaptive predictor of human mobility. Moreover, it switched between offline pre-trained and online adaptive human predictors using a gating neural network. They have adopted a real-world GPS-log dataset for training Tokyo and Osaka cities, where this approach realized a higher prediction accuracy for this approach. This framework can be employed for several purposes, for instance, incorporating additional data such as event information or weather data to predict human mobility. The framework minimizes unnecessary information by performing more than one online training simultaneously. Moreover, the used dataset is considered a little representation of the real world. However, that system is unstable due to the sparse data.

Shi et al. [ 77 ] proposed a novel model for the trajectory prediction of pedestrians in highly crowded scenarios. The model relies on using LSTM and contains a trained decoder and encoder by truncated backpropagation. The experiments used data from the trajectory train station in Tokyo, Japan. This model has proven stable concerning predictions of varying lengths. In addition, it realized an average for both Evaluation Metrics Of The Prediction Errors (Average Displacement Error And Final Displacement Error) Of 21.0%.

In the same context, [ 57 ] studied the prediction of the trajectories of foreign tourists using lstm. Nevertheless, there is a difference. The first layer of lstm is fed with the input sequence, and every other layer of lstm is fed with the layer's output that precedes it. They claimed that the proposed method outperformed classical approaches.

Zhang et al. [ 79 ] studied monitoring passenger flow in a passenger metro by creating a cnn-based platform. The proposed method has three parts: the first is a cnn group used to extract features from images. Then, the second is a module of feature extraction utilized to enhance multiscale. Finally, transposed convolution is applied to the sample to create a high-quality density map.

Lastly, some of these works used cnns with lstm methods to extract images feature in order to examine and analyze crowd scene, they have accomplished high-quality. In every case, the integration of cnns with lstm is considered an effective method to produce a high-quality density map, and thus it can give good results.

4.3.2 Crowd behavior analysis

Swathi et al. [ 37 ] developed a vigorous model, which integrates features of deep learning AlexNet (alippi, disabato and roveri, 2018) with high-dimensional features of the gray-level co-occurrence matrix (GLCM) that have hybrid deep statistical features. Moreover, it used a multi-feed forward neural network model (MFNN) to execute multi-category classification. AlexNet and GLCM provide a wealth of information on spatiotemporal features to make ideal classification decisions. The MFNN algorithm helps ideal multi-class classification. The model has achieved an accuracy rate of crowd behavior classification of 91.35%, 89.92% precision, 89.12% f-measure, and recall of 88.34%.

Zhang et al. [ 65 ] proposed a framework to predict crowd behavior in complex scenarios. The framework consists of three components: the module of scene feature extraction, the discriminator, and the generator. The first component captures the environment's visual signal, the spatial layout, and the interrelationship of pedestrians. The second component measures the similarities between the real trajectories and the generated ones. The third component consists of the encoder and the decoder parts that use lstm for inputting. Experiments are executed on the standard crowd benchmarks datasets, such as the chinese university of hong kong(cuhk) crowd (shao, change loy and wang, 2014; shao, loy and wang, 2016), the eth zurich university(eth) datasets [ 97 ], the crowd-flow, and the university of cyprus (ucy)datasets [ 98 ]. These experiments confirm that the proposed framework successfully predicts the behaviors of crowds in complex scenarios.

According to above, integration alexnet features with glcm have achieved a good accuracy rate for classification.

4.3.3 Crowd abnormality detection

Abnormal behavior is an unusual event occurring in overcrowded scenes. Therefore, crowd abnormality detection in crowded areas plays a pivotal role in preventing any disasters due to riots. The domain of anomaly detection has gained the interest of researchers in computer science in recent years.

Video anomaly detection (VAD) uses algorithms of temporal video segmentation to detect shot boundaries in sequential frames of video [ 99 ]. VAD challenges relate to crowded and complex scenes, small anomaly datasets, and anomaly localization [ 49 ]. Moreover, the challenge of false-positive detection results is that the system incorrectly discovers normal events as abnormal ones [ 49 ]. For these reasons, deep learning methods are more suitable than traditional methods [ 69 , 95 , 100 ]. In particular, unsupervised deep learning methods are the best solution [ 49 ].

Ganokratanaa et al. [ 49 ] proposed a new unsupervised deep residual spatiotemporal translation network (named DR-STN). The proposed approach has embedded with DR-cGAN and OHNM, which refer to Deep Residual conditional Generative Adversarial Network and Online Hard Negative Mining, respectively. The authors claim that their approach reduces the detection of a false-positive anomaly. Furthermore, it increases anomaly localization accuracy with a rate of 96.73%.

Wang et al. [ 38 ] proposed a novel algorithm to solve the problem of visual abnormality detection in crowd scenes. The abnormal frame is called a global abnormal event (GAE). However, determining the abnormal area in one frame is called a local abnormal event (LAE). This process uses a feature descriptor extraction of MHOFO (motion descriptor, namely a multi-frame descriptor). The motion information is represented by this descriptor after capturing it as a multi-frame. After that, captured samples are trained via a cascade deep autoencoder (CDA) as a generative network to detect abnormal behavior. Their experiment was performed on three benchmark datasets, University of California, San Diego (UCSD) [ 70 ], PETS2009 [ 73 ], and UMN [ 77 ]. They have proven that their algorithm shows competitive results. Although their model is slower than the SCL method, it is better in terms of performance. The SCL is the fastest method in the published papers for anomaly detection.

Ammar and Cherif [ 39 ] proposed a model to treat the problem of panic behavior detection in abnormal situations, which is named DeepROD. This technique worked in real time, online, and offline. It relies on statistical characterization and LTMS neural networks to predict future values of features. They claimed that their model is proven by experiments on well-known datasets (both public databases and livestreaming sources). Specifically, online training has given a better performance than offline training for the crowded scenes. Furthermore, it provided good processing time and accuracy. Nevertheless, DeepROD has lower accuracy when tested on a livestreaming source, such as a festival video.

Khan et al. [ 59 ] proposed an AlexNet-based crowd anomaly detection model to detect the anomaly in the image frame. Their model was comprised of three fully connected layers, four convolution layers, with additional the rectified linear unit (ReLU) was used as an activation function. The experiment has been performed on a personal computer using fewer computational resources, it appeared that the proposed model outperformed other studies and fulfilled 98%.

Basalamah et al. [ 50 ] proposed a Bi-LSTM framework using motion information to detect congestion rather than count pedestrians.

4.3.4 Group activity detection

Vahora et al. [ 45 ] proposed a novel model using a deep neural network for the recognition of group activity via video monitoring. The model has a multi-layer deep architecture, which integrates CNN with RNN. CNN model was used to capture information, feature, and level semantics from the scene for recognizing mysterious group activities. The RNN model used the LSTM model and gated recurrent unit (GRU) model to handle the problem of long-term dependency for the RNN model.

4.4 Social media-based analysis

Over the last few years, several smartphone social network applications (apps) have come to market. These apps enable users to exchange their information, location, and temporal data, usually called check-in data. Day by day, social network apps have become more utilized by people. Especially popular are apps based on geotagged social microblogs and location-based social networks (LBSNs) such as Twitter, Facebook, Instagram, and LinkedIn. An advantage of social media (SM) is that users share their interests and purposes when, where, and why they go out. A result is an enormous source of data that may help researchers in different domains to perform crowd analysis, such as sports, religious, and carnival events, as well as in the marketing domain and trend detections. From another perspective, the data sources of SM open other horizons in the analysis domain, including counting people, computing individual tracks, and detecting normal and abnormal behavior in a crowd [ 101 ]. Moreover, they show how much impact these data have to create a crowd or influence their behavior [ 102 ]. Figure 6 illustrates trends of analysis via social media.

Analysis trends via Social Media

4.4.1 Opinion/sentiment analysis

Öztürk and Ayvaz [ 46 ] collected all tweets in English and Turkish languages that discuss humanitarian issues concerning the Syrian refugee crisis to perform sentiment analysis on them. They used the twitter package to collect data from twitter. Then, they utilized the Rsentiment package. Both packages were developed in the r programming language. Rsentiment contains a comprehensive sentiment dictionary in English and provides a sentiment score. Whereas in the Turkish language, the authors have developed a sentiment lexicon of 5405 words. Finally, the results of these analytics, overall sentiments were positive about Syrian issues. While only 12% of the tweets in English were positive, the tweets in Turkish were equally distributed among neutral, negative, and positive sentiments. It is good to adopt this model to suit different languages.

In the same context, Malik et al. [ 64 ] created an alert system for Pakistan government authorities in order to determine the public emotions of people against upcoming anti-government.

Duan et al. [ 60 ] studied a stampede on shanghai new year’s eve in 2014. This study investigated the reasons for this crowd behavior through the viewpoint of social media data. The authors developed a framework using check-in data of the Weibo platform of three trends, the emotional fluctuations of citizens, the topic changes in posts, and the collection level of check-in data. The framework processes are executed as follows. At first, the location information of check-in data is taken from Weibo to analyze the spatial and temporal using Moran’s i index. Next, the textual data of Weibo is analyzed through topic modeling using the Latent Dirichlet Allocation (LDA) method. Finally, sentiment analysis is analyzed and divided into five groups to extract percentages of negative and positive sentiments. As a result of this study, the geographical features can directly reflect changes in crowd flow, as well as the psychological states of people before and after accidents. However, it still faced some challenges.

4.4.2 Geo-located analysis

Redondo et al. [ 48 ] proposed a hybrid solution based on clustering techniques and entropy analysis to early detect unexpected behaviors in social media. Data is collected from Location-based Social Networks (LBSNs). The authors used the Instagram platform for this study because it is a good source for geo-located data. Moreover, the APIs of some social media platforms impose limitations on the access of visual data by developers.

Finally, previous works has studied crowd flow behavior by analyzing textual data. Thus, Duan et al. [ 18 ] claimed that prior knowledge of people's psychological and behavioral states may help in understanding crowd behavior.

4.5 Crowd sound emotion recognition

Franzoni et al. [ 40 ] introduced the first model to study sound emotions for the crowd. The model integrates CNN and spectrogram-based techniques. According to their claims, they have not compared the results of their experiments with any prior study with a similar domain (crowd). However, they compared these results with studies focused on analyzing individual-speech emotions. The model has a 10% improvement in average accuracy. Their study proved that the AlexNet-CNN spectrogram-based method is appropriate to analyze the sound emotions of the crowds.

At the end of this section, this paper demonstrates the limitations of the above papers. There are some problems during the process of data extraction. The information inferred was insufficient due to a lack of data on procedure, methods, and performance, which may be reflected by the QA.

4.6 Crowd management at Hajj event

This part will display all studies that support Hajj research. Hajj ritual is the fifth pillar of Islamic, every Muslim should visit to the holy places in Makkah, Saudi Arabia once at least in his life. They should able financially and physically to perform worships of Hajj. The period of Hajj is between 8 and 12th of the 12th month every year, it is called (Dhulhijjah) in the Islamic (lunar) calendar. Hajj's crowd of up to three million people comprised of pilgrims from all over the world in one sacred spot. The geographic area of the holy sites for performing Hajj rituals is not exceeding 33 km 2 . This makes the Hajj authorities to face a great challenge to deal with the overcrowding of the Hajj in a specific period and place, firstly relating to the security and safety of pilgrims. The objective of our work is to discover the efficient practices of crowd management in the holy biggest event in Saudi Arabia, it is the Hajj ritual [ 103 , 104 ]. All methods currently applied in the field of Hajj crowd management are still lack of attention and development from researchers, especially in terms of exploiting of textual data to analysis of crowd's emotions, sentiments, or opinions. The two papers are picked below according to our criteria in this survey.

Farooq et al. [ 41 ] presented a novel method for abnormal behavior detection for crowds that may lead to dangerous disasters, such as a stampede. The model captures motion in the form of images, then classifies these images according to crowd divergence behavior using a CNN method, where the CNN has been trained on motion-shape images (MSIs). Moreover, the finite-time Lyapunov exponent (FTLE) domain is acquired when the optical flow (OPF) is computed first. LCS (Lagrangian coherent structure) in the FTLE domain represents dominant motion for the crowd. Finally, a scheme of ridge extraction transforms the LCS-to-grayscale MSIs. The model is tested on six real-world low and high-density datasets. They claimed that the experiments produced effective results for their method in terms of detecting divergence accurately, as well as detecting starting points of congestion at high and at low density. Furthermore, they presented two new datasets, including video scenes of normal and abnormal behaviors for a high-density crowd. In the Hajj case, the authors have applied their model to pilgrims’ crowds at Makkah, Saudi Arabia. It used recorded Video data (the PILGRIM dataset) taken from a live broadcast of the Makkah TV channel. They have generated three behavior videos from every single video. The proposed method also has outperformed this dataset.

Habib et al. [ 61 ] developed a novel framework to identify abnormal activity for pilgrims at Makkah. A lightweight CNN model was trained on the dataset of pilgrims. This dataset was captured from installed CCTV et al.-Haram. The images’ frames were passed to the proposed model for the extraction of spatial features. Then, an LSTM network was created for the extraction of temporal features. Lastly, the system will make an alarm when an emergency occurs, such as an accident or violent activity, to inform the authorities to take the appropriate action. They have performed experiments on two violent activity datasets: Hockey Fight and Surveillance Fight. The model achieved good accuracies of 98.00 and 81.05, respectively. However, this model suffers from the shortcoming of recognizing violent activity from one perspective only. And it is the best that recognizes violent activity from multiple perspectives to obtain insights into the activities.

Finally, the review concludes from Sects. 4 and 5 that most of these studies have been concerned with CNN methodology and integrated with other techniques to benefit further and improve the accuracy of model performance, Table 6 summarizes all studies that used CNN approach with their advantages and disadvantages. Table 7 displays that used the methodologies of RNN such as LSTM approach. While Table 8 illustrates studies that used different methodologies to create models.

According to our viewpoint, the models of crowd counting still need development for several problems as follows: Detecting large objects as people, but do not detect small objects. Most of models cannot be generalized on all datasets, where they give good results with some datasets and inefficient results with others. Classifying some objects as people by mistake. The accuracy of captured image/video varies according to the installed camera angles. Hence, the challenges can be minimized as possible, installing cameras at every angle in the place to ensure the monitoring of all people and improving the accuracy of image/video pixels for the extraction of features efficiently. At last, internet of things (IoT) devices can improve counting crowds besides DL models. In terms of analysis of social media, geographic locations feature should be exploited for processes analytic of crowds. Furthermore, the various languages should be supported just like the English language. Consideration of the backgrounds of psychological, social, and beliefs of people when studying their expression on social media.

5 Results analysis of comprehensive study of crowd management

In this study, the comprehensive study is divided into two domains. The first one is crowd management in various events in the world. The second one is crowd management in local area named Hajj event in Makkah, Saudia Arabia. All these studies focus on approaches that applied DL techniques. In terms of comprehensive study of crowd management in various events, the related works are divided into four sections, everyone has two or more subsections. The sections are crowd statistics, crowd scene analysis, social media-based analysis, and crowd sound emotion recognition.

5.1 Crowd statistics

Many new hybrid approaches have been developed by CNN-based methods to improve crowd counting and crowd density estimation fields. The outcomes of combining two or more methods have confirmed that hybrid techniques enhance performance, increase accuracy, and prevent many obstacles in computer vision projects. This has been displayed using the adaptive convolutional neural network (ACNN)-based model, locality-aware features (LAF) integrated with CNN features, multi-column convolutional neural network (MCNN) method, two-task convolutional neural network (T2CNN) with dense degree classification (DDC), and end-to-end semantic scene segmentation (SSS) based on CNN for calculation of crowd counting and density.

5.2 Crowd scene analysis

Many works have attempted to find the successful solutions for normal or abnormal human behavior analysis. Videos scene analysis for human behavior detection is faced many challenges relate to crowded and complex scenes, anomaly localization, small anomaly datasets, and false-positive detection [ 56 ]. To solve these reasons, DL techniques are more suitable and successful solutions than traditional methods to treat problems and challenges. The researches have been developed by CNN-based methods integrated with other techniques. The works of Shi et al. [ 50 ] and Crivellari and Beinat [ 51 ] have used backpropagation processing through the LSTM technique. LSTM in the first layer feeds subsequent layers and every other layer of LSTM is provided with the layer's output that precedes it. Furthermore, the model of Zhang et al. [ 54 ] used LSTM in third component for inputting stage to improve accuracy rate for classification. Moreover, Ammar and Cherif [ 61 ] have integrated LTMS neural networks and statistical characterization to predict future values of features. Vahora et al. [ 31 ] used LSTM beside GRU to handle the problem of long-term dependency, also used CNN to capture features from the scenes.

Finally, some of these works used CNNs with RNN methods such as LSTM and GRU to treat long-term dependency problems, increase improve extraction of scene features, and accomplish high-quality to anomaly behavior detection.

5.3 Social media-based analysis

Some of the works focus on studying crowd management through text data analysis is less compared to analyzing image and video scenes. Due to of the difficulties associated with natural language processing that make it difficult to understand the intentions of people's feelings and emotions towards events and situations. Moreover, It would be excellent if there is prior knowledge of people's social, psychological and behavioral states in order to understand crowd behavior to prevent emergency cases proactively to ensure crowd safety [ 43 ].

5.4 Crowd sound emotion recognition

There is one research that has presented to study emotion detection by sounds, where it used AlexNet-CNN spectrogram-based method to analyze the sound emotions of the crowds. Spectrogram-based method is appropriate for crowd sounds analysis.

5.5 Crowd management at Hajj event

In terms of crowd management at hajj event, the flow of data during a Hajj period is huge, whether it is visual, text, or audio data. This digital wealth must be greatly exploited by the Hajj authorities to reduce the terrible effects that may be when proactive solutions are not developed to control crowds. Like previous studies, CNN with LSTM have been used to effectively extract visual attributes to classify anomalous or anomalous crowd behavior, this achieved excellent accuracy results.

5.6 Addressing of used methodologies limitations

It is clear from the review of existing works that crowd management is plagued by drawbacks and challenges that have restricted its rapid improvement in recent years. Hence, we made several important considerations when building crowd models.

It can be noticed most of the works of literature is that they still remain density-dependent. It means their models developed for macro-analysis independently from micro-analysis, while applications of real-world require crowd analysis to be conducted starting at macro-level and branching down into the micro-level. Therefore, it is important in future works in terms of modelling surveillance, behavioral understanding of crowds, must concern on the enhancement at both macro-and micro-levels and integrated between them.

Furthermore, it can be observed that most of the literature based on computer vision is performed under strong and restrictive conditions, for example, the perspective of the installed camera in the place, surrounding environment, estimating density of crowds, noise, etc. It is vital to realize that these requirements are inherited from the computer vision field since they are viewed as extension techniques for crowd modeling. There is a common sense of acceptance of these challenges for researchers. Whereas in this work, we recommend the integration of some techniques that collect data about people to reduce absolute dependence on traditional equipment. We claim that utilizing various data on social media will make a vital source for video surveillance domain and counting crowd density, behavior, or abnormality crowd detection, which increases the cognitive diversity and learning new patterns for datasets. Hence, the crowd models will be generalizability on the different environments. It has increased the number of large-scale events in the world; thus, the organizers should benefit from the deepest insights about attendees’ characteristics besides events' characteristics. It becomes possible to describe the behavior of people during crowd events using social media data, this is paving the path for crowd monitoring and management by using real-time applications [ 105 ]. The work of [ 106 ] is a good example of exploiting the data in streaming channels and social media during the Hajj season. We believe that, in the future, social media data related to expressing people’s daily lives will become close to understanding the behavior of crowds during events.

The future research must be concerned with the complementarity of the models to solve challenges and drawbacks rather than with minor developments to increase the accuracy of the model only. For these reasons, it is important to understand the differences between the kinds of supervised and unsupervised DL techniques. Many of the relevant concepts may confused together when building large or complex models. Table 9 clears the most important strengths and weaknesses of ML and DL algorithms.

One of the main problems with the majority of local works during Hajj season is that most research is performed in isolation as urban planning for smart cities and the variety of needs regarding crowd management. Urban crowd management is an integral activity for any event, such as crowd flow, estimating density, monitoring street grid, movement of buses [ 107 ], crowd trajectory [ 108 ], impact of a pandemic on crowds [ 109 ], etc. Integration Urban provides good decision support for the development of the city in all respects. Big data and advanced intelligence computational techniques can help the planning, design, management, analysis, and simulation of smart cities. For instance, early planning of safety evacuations in the midst of a natural disaster incident based on location data of mobile phones leveraging both machine learning approaches [ 110 ]. Use crowd-harvested data to study the population's sentiments, traffic patterns, and perceptions of neighborhoods, in addition, to simulating the model urban systems more realistically, which is crowdsourcing effective the analyzing and modeling of urban morphology at much finer social scales, temporal, and spatial [ 111 ]. Hence, smart cities will be able to control and monitor dynamic changes as they happen inside the city during crowded events. For instance, using the FOPID controllers controls systems with nonlinear dynamics, also improves the complex systems performance in various applications [ 112 ], and using DETDO optimizer to solve real-world engineering design problems [ 113 ].

6 Discussion

In this SLR, about 45 DL-based articles are reviewed. According to the detailed analysis of various crowd management approaches and their state-of-the-art performance in this survey, our survey forecasts that DL-based methods will predominate future research in the crowd analysis and management fields. It can be noticed that most methods have been integrated with other DL methods, such as CNN with LSTM, to increase the accuracy performance. Moreover, variations in types of input, layers, or fed to in the CNN. It utilizes popular datasets or creates a novel dataset for performing testing on proposed approaches.

Most of these papers focused on crowd scene analysis in the computer vision field. Therefore, the major challenge for crowd management is the lack of sentiment analysis of crowd-based big data on social media. There is also a lack of custom datasets to feed textual data analysis. Owing to the challenges related to natural language processing, it makes difficult to understand people's emotions towards events and situations by textual expression. Furthermore, studying of people's psychological and behavioral it may reduce the severity of these challenges. Furthermore, its open scope for a greater understanding of what is behind the meanings and words. Thus, investigation of the crowd’s behavior from all aspects is greatly crucial for crowd safety, also to prevent dangerous emergency situations before they happened [ 43 ].

6.1 Current study vision discussion

This section discusses our vision of this study compared to previous state-of-the-art studies. The main objective of this survey is to spotlight the shortcomings or defects of previous papers. New approaches must provide effective solutions for crowd video analysis in real-time, while traditional approaches are not able to handle efficient solutions in a time-bounded manner. Traditional approaches are insufficient for crowd analysis cause the size of the crowd is huge and dynamic in real-world scenarios. In addition, the behavior and actions of individuals are difficult to identify. The shortcomings can be identified in existing approaches as follows: real-world dynamics, time complexity, bad weather conditions, overlapping of objects [ 119 ], and unexpected incidents. All existing approaches were handling the shortcomings independently. It can be observed that there seem to be almost no concerns about the lack of research in sentiment analysis of crowd-based big data on social media in the world, especially during Hajj events. It can be perceived as a missed chance to learn from different visions. Thus, this study seeks to change imbalance research by configuring new Integrative frameworks and methodologies and highlighting the prior good practices in this domain. It is significant that the researcher community realizes these gaps when constructing existing systems and continuing to monitor the development of integrated research in crowd management.

This paper aims to support Hajj research through the enhancement of the pilgrims' behavior analysis and work to cover the above aspects. Moreover, we will provide datasets of pilgrims taken from social media and will make them publicly available to be useful to other researchers. The authors are seeking “actionable SM-based crowd management”. In this sense, traditional crowd management needs a new multidimensional conception. To build a new robust infrastructure, it must be integrated as follows; “AI algorithm + computing power + big data = smart service” [ 120 ]. It is significant that utilize the best optimizers to improve the performance of complex systems such as FOPID [ 112 ], DETDO [ 113 ], Genghis Khan shark [ 121 ], Geyser Inspired Algorithm [ 122 ], Prairie Dog Optimization Algorithm [ 123 ], Dwarf Mongoose Optimization Algorithm [ 124 ], and Gazelle Optimization Algorithm [ 125 ].

The advantage of this work is the crowd management domain may rise to another sophisticated level if considering the attention on social media big data. Presenting a new taxonomy of crowd management based on deep learning algorithms, including all domains of analyzing data: visual, audio, and textual. Presenting the comprehensive examination of the global crowd management works in order to benefit Hajj authorities to apply the best practices locally. The following research [ 53 , 126 ] can be contributed to addressing weaknesses in Hajj research regarding counting crowds, where the model focuses on the attention of CNN channel-wise, spatial-wise attention, and context-aware. In addition, the model can be used for other purposes, such as image supporter recognition, localization of individuals, and face recognition. Furthermore in [ 46 ], the authors provide a good model for the analysis of textual data and understanding the emotions of users, but it needs some modifications to suit other languages like Arabic.

There are some lessons that can be learnt from this SLR. First, our practice of SLR has emphasized the lack of Hajj data analysis topics. SLR employment is valuable to stay informed about those topics to support the data or knowledge of Hajj. Second, this SLR discovered that the future direction of data analysis and prediction depends on the development of CNN models. Third, crowd analysis studies focus on the analysis aspect of video surveillance more than textual data.

Limitations of this SLR include intentionally ignoring conference papers because they contain incomplete models or studies. Thus, we limited our sources to academic articles only. During data extraction, we found insufficient information related to the environment, evaluation, accuracy, and procedure in some papers, which may be reflected by the QA. As a result, some inferred data may have inaccuracies due to unclear information in the papers. Table 10 shows the summarization of the learning lesson, limitations, and future research directions.

7 Conclusion

In this SLR, the researchers explored comprehensive crowd management from the aspect of DL methods. This survey has performed a wide investigation for relevant related works published in the interval 2010 to 2023. Moreover, the survey elicited pivotal information based on our research questions (RQs). The research goals have been fulfilled effectively through these RQs that were established to examine and analyze the scope of the research. Moreover, ensuring that the key findings and contributions have been performed usefulness for future researchers, also the gaps and obstacles that faced crowd management have been discussed in the Sects. 5.6 and 6 . The four RQs raised in this SLR, and their findings are as follows:

RQ1: Most previous works have been classified into two categories, crowd scene analysis and crowd statistics. However, these previous works omit the opinion mining of users on social media to predict future crowd actions.

RQ2: Supervised and unsupervised DL methods provide high accuracy in general, which supported computer vision in crowd analysis in many studies, especially in architectures based on the CNN model. Because of its high efficiency and accuracy, it has become the most reliability model for researchers.

RQ3: We observed there is the fewest number of studies regarding crowd management at Hajj. As well they focused on the detection of abnormalities in crowded scenes.

The primary aim of our review is to investigate crowd analysis fields in every gathering around the world. Furthermore, the local gathering, especially crowd analysis in a Hajj event for example. The paper aims to illustrate the dilemmas and obstacles that have been faced in every study. Moreover, it aims to find research gaps existing to focus on it in future studies. Finally, we observe that most studies were about crowd scene analysis via private/publicly available datasets or live-streaming surveillance, whether using supervised or unsupervised techniques. They ignored the behavior analytics and predicted it by textual data on social media. In addition, the literature indicates the lack of Hajj research, especially in sentiment analysis and the study of the pilgrims' behavior. Overall, the systematic literature review links the widespread knowledge transfer debate of crowd management in terms of the study behavior of entities via SM big data and predicting the actions. Thus, the current study enriches the research communities and academic discourse.

Availability of data and materials

Availability upon request from the Corresponding author.

Tripathi G, Singh K, Vishwakarma DK (2019) Convolutional neural networks for crowd behaviour analysis: a survey. Vis Comput 35(5):753–776

Article Google Scholar

Alghamdi N, Alageeli N, Sharkh DA, Alqahtani M, Al-Razgan M (2020) An eye on Riyadh tourist season: using geo-tagged snapchat posts to analyse tourists impression. In: 2020 2nd International conference on computer and information sciences (ICCIS), pp 1–6

Moore BE, Ali S, Mehran R, Shah M (2011) Visual crowd surveillance through a hydrodynamics lens. Commun ACM 54(12):64–73

Kizrak MA, Bolat B (2021) Crowd density estimation by using attention based capsule network and multi-column CNN. IEEE Access 9:75435–75445

Zafarani R, Abbasi MA, Liu H (2014) Social media mining: an introduction. Cambridge University Press, Cambridge

Book Google Scholar

Tawfik GM et al (2019) A step by step guide for conducting a systematic review and meta-analysis with simulation data. Trop Med Health 47:1–9

Hiebl MRW (2023) Sample selection in systematic literature reviews of management research. Organ Res Methods 26(2):229–261

Kitchenham B, Brereton OP, Budgen D, Turner M, Bailey J, Linkman S (2009) Systematic literature reviews in software engineering—a systematic literature review. Inf Softw Technol 51(1):7–15

Vashishth TK, Kumar B, Sharma V, Chaudhary S, Kumar S, Sharma KK (2023) The evolution of AI and its transformative effects on computing: a comparative analysis. In: Mishra BK (ed) Intelligent engineering applications and applied sciences for sustainability. IGI Global, Hershey, pp 425–442

Chapter Google Scholar

Dybå T, Dingsøyr T (2008) Empirical studies of agile software development: a systematic review. Inf Softw Technol 50(9–10):833–859

Ofem P, Isong B, Lugayizi F (2022) On the concept of transparency: a systematic literature review. IEEE Access 10:89887–89914

Sánchez FL, Hupont I, Tabik S, Herrera F (2020) Revisiting crowd behaviour analysis through deep learning: taxonomy, anomaly detection, crowd emotions, datasets, opportunities and prospects. Inf Fusion 64:318–335

Kraft T, Wang DX, Delawder J, Dou W, Yu L, Ribarsky W (2013) Less after-the-fact: investigative visual analysis of events from streaming twitter. In: 2013 IEEE symposium on large-scale data analysis and visualization (LDAV), pp 95–103

Yogameena B, Nagananthini C (2017) Computer vision based crowd disaster avoidance system: a survey. Int J Disaster Risk Reduct 22:95–129

Zhang X, Yu Q, Yu H (2018) Physics inspired methods for crowd video surveillance and analysis: a survey. IEEE Access 6:66816–66830

Bendali-Braham M, Weber J, Forestier G, Idoumghar L, Muller P-A (2021) Recent trends in crowd analysis: a review. Mach Learn Appl 4:100023

Google Scholar

Son LH, Pritam N, Khari M, Kumar R, Phuong PTM, Thong PH (2019) Empirical study of software defect prediction: a systematic mapping. Symmetry (Basel) 11(2):212

Khari M, Kumar P (2019) An extensive evaluation of search-based software testing: a review. Soft Comput 23(6):1933–1946

Agarwal A, Singh R, Khari M (2022) Detection of DDOS attack using IDS mechanism: a review. In: 2022 1st International conference on informatics (ICI), pp 36–46

Merugu S, Tiwari A, Sharma SK (2021) Spatial–spectral image classification with edge preserving method. J Indian Soc Remote Sens 49(3):703–711

Yang S, Linares-Barranco B, Chen B (2022) Heterogeneous ensemble-based spike-driven few-shot online learning. Front Neurosci. https://doi.org/10.3389/fnins.2022.850932

Yang S, Tan J, Chen B (2022) Robust spike-based continual meta-learning improved by restricted minimum error entropy criterion. Entropy 24(4):455

Article MathSciNet Google Scholar

Yang S et al (2022) SAM: a unified self-adaptive multicompartmental spiking neuron model for learning with working memory. Front Neurosci. https://doi.org/10.3389/fnins.2022.850945

Yang S, Gao T, Wang J, Deng B, Lansdell B, Linares-Barranco B (2021) Efficient spike-driven learning with dendritic event-based processing. Front Neurosci 15:601109

Najafabadi MM, Villanustre F, Khoshgoftaar TM, Seliya N, Wald R, Muharemagic E (2015) Deep learning applications and challenges in big data analytics. J Big Data 2:1–21

Grant JM, Flynn PJ (2017) Crowd scene understanding from video: a survey. ACM Trans Multimed Comput Commun Appl 13(2):1–23

Kang D, Dhar D, Chan AB (2016) Crowd counting by adapting convolutional neural networks with side information. arXiv Prepr. arXiv1611.06748

Marsden M, McGuinness K, Little S, O’Connor NE (2016) Fully convolutional crowd counting on highly congested scenes. arXiv Prepr. arXiv1612.00220

Kumagai S, Hotta K, Kurita T (2017) Mixture of counting cnns: adaptive integration of cnns specialized to specific appearance for crowd counting. arXiv Prepr. arXiv1703.09393

Gong VX, Daamen W, Bozzon A, Hoogendoorn SP (2021) Counting people in the crowd using social media images for crowd management in city events. Transportation (Amst) 48(6):3085–3119

Huang J, Di X, Wu J, Chen A (2020) A novel convolutional neural network method for crowd counting. Front Inf Technol Electron Eng 21(8):1150–1160

Jiang X et al (2022) Transferring priors from virtual data for crowd counting in real world. Front Comput Sci 16(3):1–8

Zhang L, Yan L, Zhang M, Lu J (2021) T 2 CNN: a novel method for crowd counting via two-task convolutional neural network. Vis Comput 39(1):73–85

Shang C, Ai H, Yang Y (2019) Crowd counting via learning perspective for multi-scale multi-view web images. Front Comput Sci 13(3):579–587

Jiang H, Jin W (2019) Effective use of convolutional neural networks and diverse deep supervision for better crowd counting. Appl Intell 49(7):2415–2433

Khan SD, Basalamah S (2021) Scale and density invariant head detection deep model for crowd counting in pedestrian crowds. Vis Comput 37(8):2127–2137

Swathi HY, Shivakumar G (2021) Hybrid feature-assisted neural model for crowd behavior analysis. SN Comput Sci 2(4):1–11

Wang T, Qiao M, Zhu A, Shan G, Snoussi H (2020) Abnormal event detection via the analysis of multi-frame optical flow information. Front Comput Sci 14(2):304–313

Ammar H, Cherif A (2021) DeepROD: a deep learning approach for real-time and online detection of a panic behavior in human crowds. Mach Vis Appl 32(3):1–15

Franzoni V, Biondi G, Milani A (2020) Emotional sounds of crowds: spectrogram-based analysis using deep learning. Multimed Tools Appl 79(47):36063–36075

Farooq MU, Saad MNM, Khan SD (2021) Motion-shape-based deep learning approach for divergence behavior detection in high-density crowd. Vis Comput 38(5):1553–1577

Li H, Zhang S, Kong W (2020) Bilateral counting network for single-image object counting. Vis Comput 36(8):1693–1704

Bansal H, Sharma K, Khari M (2022) Crowd analytics: literature and technological assessment. Multimed Tools Appl 81(11):15249–15283

Fu M, Xu P, Li X, Liu Q, Ye M, Zhu C (2015) Fast crowd density estimation with convolutional neural networks. Eng Appl Artif Intell 43:81–88

Vahora SA, Chauhan NC (2019) Deep neural network model for group activity recognition using contextual relationship. Eng Sci Technol Int J 22(1):47–54

Öztürk N, Ayvaz S (2018) Sentiment analysis on twitter: a text mining approach to the Syrian refugee crisis. Telemat Inf 35(1):136–147

Hu Y, Chang H, Nian F, Wang Y, Li T (2016) Dense crowd counting from still images with convolutional neural networks. J Vis Commun Image Represent 38:530–539

Redondo RPD, Garcia-Rubio C, Vilas AF, Campo C, Rodriguez-Carrion A (2020) A hybrid analysis of LBSN data to early detect anomalies in crowd dynamics. Futur Gener Comput Syst 109:83–94

Ganokratanaa T, Aramvith S, Sebe N (2021) Video anomaly detection using deep residual-spatiotemporal translation network. Pattern Recogn Lett 155:143–150

Basalamah S, Khan SD, Felemban E, Naseer A, Rehman FU (2023) Deep learning framework for congestion detection at public places via learning from synthetic data. J King Saud Univ Inf Sci 35(1):102–114

Carvalho J, Marques M, Costeira JP (2017) Understanding people flow in transportation hubs. IEEE Trans Intell Transp Syst 19(10):3282–3291

Liu Z, Chen Y, Chen B, Zhu L, Wu D, Shen G (2019) Crowd counting method based on convolutional neural network with global density feature. IEEE Access 7:88789–88798

Elharrouss O, Almaadeed N, Abualsaud K, Al-Maadeed S, Al-Ali A, Mohamed A (2022) FSC-set: counting, localization of football supporters crowd in the stadiums. IEEE Access 10:10445–10459

Sheng B, Shen C, Lin G, Li J, Yang W, Sun C (2016) Crowd counting via weighted VLAD on a dense attribute feature map. IEEE Trans Circuits Syst Video Technol 28(8):1788–1797

Khan K et al (2021) Crowd counting using end-to-end semantic image segmentation. Electronics 10(11):1293

Shi X, Shao X, Guo Z, Wu G, Zhang H, Shibasaki R (2019) Pedestrian trajectory prediction in extremely crowded scenarios. Sensors 19(5):1223

Crivellari A, Beinat E (2020) LSTM-based deep learning model for predicting individual mobility traces of short-term foreign tourists. Sustainability 12(1):349

Zhang J, Liu J, Wang Z (2021) Convolutional neural network for crowd counting on metro platforms. Symmetry (Basel) 13(4):703

Khan AA et al (2022) Crowd anomaly detection in video frames using fine-tuned AlexNet model. Electronics 11(19):3105

Duan J, Zhai W, Cheng C (2020) Crowd detection in mass gatherings based on social media data: a case study of the 2014 shanghai new year’s eve stampede. Int J Environ Res Public Health 17(22):8640

Habib S et al (2021) Abnormal activity recognition from surveillance videos using convolutional neural network. Sensors 21(24):8291

Ebrahimpour Z, Wan W, Cervantes O, Luo T, Ullah H (2019) Comparison of main approaches for extracting behavior features from crowd flow analysis. ISPRS Int J Geo-Inf 8(10):440

He D et al (2023) A spatio-temporal hybrid neural network for crowd flow prediction in key urban areas. Electronics 12(10):2255

Malik T et al (2023) Crowd control, planning, and prediction using sentiment analysis: an alert system for city authorities. Appl Sci 13(3):1592

Zhang B, Zhang R, Bisagno N, Conci N, De Natale FGB, Liu H (2021) Where are they going? predicting human behaviors in crowded scenes. ACM Trans Multimed Comput Commun Appl 17(4):1–19

Fan Z, Song X, Xia T, Jiang R, Shibasaki R, Sakuramachi R (2018) Online deep ensemble learning for predicting citywide human mobility. Proc ACM Interact Mob Wearable Ubiquit Technol 2(3):1–21

Zou Z, Shao H, Qu X, Wei W, Zhou P (2019) Enhanced 3D convolutional networks for crowd counting. arXiv Prepr. arXiv1908.04121

Du Z, Shi M, Deng J, Zafeiriou S (2022) Redesigning multi-scale neural network for crowd counting. arXiv Prepr. arXiv2208.02894

Zhou JT, Du J, Zhu H, Peng X, Liu Y, Goh RSM (2019) Anomalynet: an anomaly detection network for video surveillance. IEEE Trans Inf Forensics Secur 14(10):2537–2550

Wang Z, Bovik AC, Sheikh HR, Simoncelli EP (2004) Image quality assessment: from error visibility to structural similarity. IEEE Trans image Process 13(4):600–612

Zhang C, Li H, Wang X, Yang X (2015) Cross-scene crowd counting via deep convolutional neural networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 833–841

Chen K, Loy CC, Gong S, Xiang T (2012) Feature mining for localised crowd counting. BMVC 1(2):3

Ferryman J, Shahrokni A (2009) Pets2009: dataset and challenge. In: 2009 Twelfth IEEE international workshop on performance evaluation of tracking and surveillance, pp 1–6

Shao J, Change Loy C, Wang X (2014) Scene-independent group profiling in crowd. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 2219–2226

Shao J, Loy CC, Wang X (2016) Learning scene-independent group descriptors for crowd understanding. IEEE Trans Circuits Syst Video Technol 27(6):1290–1303

Zhang Y, Zhou D, Chen S, Gao S, Ma Y (2016) Single-image crowd counting via multi-column convolutional neural network. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 589–597

Bird N, Atev S, Caramelli N, Martin R, Masoud O, Papanikolopoulos N (2006) Real time, online detection of abandoned objects in public areas. In: Proceedings 2006 IEEE international conference on robotics and automation, 2006. ICRA 2006, pp 3775–3780

Maalouf M, Trafalis TB (2011) Rare events and imbalanced datasets: an overview. Int J Data Min Model Manage 3(4):375–388

Ali MI, Gao F, Mileo A (2015) Citybench: a configurable benchmark to evaluate rsp engines using smart city datasets. In: International semantic web conference, pp 374–389

Li T, Chang H, Wang M, Ni B, Hong R, Yan S (2014) Crowded scene analysis: a survey. IEEE Trans Circuits Syst Video Technol 25(3):367–386

Kiran BR, Thomas DM, Parakkal R (2018) An overview of deep learning based methods for unsupervised and semi-supervised anomaly detection in videos. J Imaging 4(2):36

Al-Shaery AM, Alshehri SS, Farooqi NS, Khozium MO (2020) In-depth survey to detect, monitor and manage crowd. IEEE Access 8:209008–209019

Dalal R, Khari M, Anzola JP, García-Díaz V (2021) Proliferation of opportunistic routing: a systematic review. IEEE Access 10:5855–5883

Dai G, Hu X, Ge Y, Ning Z, Liu Y (2021) Attention based simplified deep residual network for citywide crowd flows prediction. Front Comput Sci 15(2):1–12

Festivalgoers Killed in Stampede at Love Parade in Germany (2024) [Online]. https://www.theguardian.com/world/2010/jul/24/love-parade-festival-tunnel-stampede .

S.A. Ministry of Health 2015 Health services for Hajj and Umrah general department: strenuous efforts and important roles reflecting the MOH’s readiness (2024) [Online]. https://www.moh.gov.sa/en/Ministry/MediaCenter/News/Pages/News-%0A2015-09-24-002.aspx

Dollar P, Wojek C, Schiele B, Perona P (2011) Pedestrian detection: an evaluation of the state of the art. IEEE Trans Pattern Anal Mach Intell 34(4):743–761

Chan AB, Vasconcelos N (2011) Counting people with low-level features and Bayesian regression. IEEE Trans image Process 21(4):2160–2177

Tang NC, Lin Y-Y, Weng M-F, Liao H-YM (2014) Cross-camera knowledge transfer for multiview people counting. IEEE Trans Image Process 24(1):80–93

“ucf-cc-50 Dataset” (2024) [Online]. http://crcv.ucf.edu/da-ta/ucf-cc-50/ .

Xu Z, Lin H, Chen Y, Li Y (2024) Label noise robust crowd counting with loss filtering factor. Appl Artif Intell 38(1):2329859

Idrees H et al. (2018) Composition loss for counting, density map estimation and localization in dense crowds. In: Proceedings of the European conference on computer vision (ECCV), pp 532–546

Liu L, Chen J, Wu H, Li G, Li C, Lin L (2021) Cross-modal collaborative representation learning and a large-scale rgbt benchmark for crowd counting. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 4823–4833

Wang Q, Gao J, Lin W, Li X (2020) NWPU-crowd: a large-scale benchmark for crowd counting and localization. IEEE Trans Pattern Anal Mach Intell 43(6):2141–2149

Fan Y, Wen G, Li D, Qiu S, Levine MD, Xiao F (2020) Video anomaly detection and localization via Gaussian mixture fully convolutional variational autoencoder. Comput Vis Image Underst 195:102920

Alippi C, Disabato S, Roveri M (2018) Moving convolutional neural networks to embedded systems: the alexnet and VGG-16 case. In: 2018 17th ACM/IEEE International conference on information processing in sensor networks (IPSN), pp 212–223

Pellegrini S, Ess A, Schindler K, Van Gool L (2009) You’ll never walk alone: Modeling social behavior for multi-target tracking. In: 2009 IEEE 12th International conference on computer vision, pp 261–268

Lerner A, Chrysanthou Y, Lischinski D (2007) Crowds by example. Comput Graph Forum 26(3):655–664

Abdulhussain SH, Al-Haddad SAR, Saripan MI, Mahmmod BM, Hussien A (2020) Fast temporal video segmentation based on krawtchouk-tchebichef moments. IEEE Access 8:72347–72359

Ganokratanaa T, Aramvith S, Sebe N (2020) Unsupervised anomaly detection and localization based on deep spatiotemporal translation network. IEEE Access 8:50312–50329

Riboni D, Bettini C (2015) Incremental release of differentially-private check-in data. Pervasive Mob Comput 16:220–238

Stieglitz S, Mirbabaie M, Ross B, Neuberger C (2018) Social media analytics—challenges in topic discovery, data collection, and data preparation. Int J Inf Manage 39:156–168

Owaidah AA, Olaru D, Bennamoun M, Sohel F, Khan RN (2021) Modelling mass crowd using discrete event simulation: a case study of integrated Tawaf and Sayee rituals during Hajj. IEEE Access 9:79424–79448

Al-Nabhan N et al (2021) An intelligent IoT approach for analyzing and managing crowds. IEEE Access 9:104874–104886

Gong VX, Daamen W, Bozzon A, Hoogendoorn SP (2020) Crowd characterization for crowd management using social media data in city events. Travel Behav Soc 20(March):192–212. https://doi.org/10.1016/j.tbs.2020.03.011

Shambour MK (2022) Analyzing perceptions of a global event using CNN-LSTM deep learning approach: the case of Hajj 1442 (2021). PeerJ Comput Sci 8:e1087

Felemban E, Rehman FU, Biabani AA, Naseer A, Hussain O, Warriach EU (2020) An interactive system for analyzing movement of buses in Hajj. J Theor Appl Inf Technol 98(21):3468–3481

Felemban E, Khan SD, Naseer A, UrRehman F, Basalamah S (2021) Deep trajectory classification model for congestion detection in human crowds. Comput Mater Contin 68(1):705–725

Gutub A, Shambour MK, Abu-Hashem MA (2023) Coronavirus impact on human feelings during 2021 Hajj season via deep learning critical Twitter analysis. J Eng Res 11(1):100001

Yin L et al (2020) Improving emergency evacuation planning with mobile phone location data. Environ Plan B Urban Anal City Sci 47(6):964–980

Crooks A et al (2015) Crowdsourcing urban form and function. Int J Geogr Inf Sci 29(5):720–741

Abualigah L, Ekinci S, Izci D, Zitar RA (2023) Modified elite opposition-based artificial hummingbird algorithm for designing FOPID controlled cruise control system. Intell Autom Soft Comput 38(2):169–183

Hu G, Zheng Y, Abualigah L, Hussien AG (2023) DETDO: An adaptive hybrid dandelion optimizer for engineering optimization. Adv Eng Inf 57:102004

Karamizadeh S, Abdullah SM, Halimi M, Shayan J, Javad Rajabi M (2014) Advantage and drawback of support vector machine functionality. In: 2014 International conference on computer, communications, and control technology (I4CT), pp 63–65

Somvanshi M, Chavan P, Tambade S, Shinde SV (2016) A review of machine learning techniques using decision tree and support vector machine. In: 2016 International conference on computing communication control and automation (ICCUBEA), pp 1–7

Cox DR, Snell EJ (2018) Analysis of binary data. Routledge, Milton Park

Aggarwal CC, Zhai C (2012) A survey of text classification algorithms. In: Aggarwal CC, Zhai C (eds) Mining text data. Springer, Boston, pp 163–222

Liu Y, Wang Y, Zhang J (2012) New machine learning algorithm: random forest. In: Information computing and applications: third international conference, ICICA 2012, Chengde, China, September 14–16, 2012. Proceedings, vol 3, pp 246–252

Sreenu G, Durai S (2019) Intelligent video surveillance: a review through deep learning techniques for crowd analysis. J Big Data 6(1):1–27

Xue X et al (2022) Research roadmap of service ecosystems: a crowd intelligence perspective. Int J Crowd Sci 6(4):195–222

Hu G, Guo Y, Wei G, Abualigah L (2023) Genghis Khan shark optimizer: a novel nature-inspired algorithm for engineering optimization. Adv Eng Inf 58:102210

Ghasemi M, Zare M, Zahedi A, Akbari M-A, Mirjalili S, Abualigah L (2024) Geyser inspired algorithm: a new geological-inspired meta-heuristic for real-parameter and constrained engineering optimization. J Bionic Eng 21(1):374–408

Ezugwu AE, Agushaka JO, Abualigah L, Mirjalili S, Gandomi AH (2022) Prairie dog optimization algorithm. Neural Comput Appl 34(22):20017–20065

Agushaka JO, Ezugwu AE, Abualigah L (2022) Dwarf mongoose optimization algorithm. Comput Methods Appl Mech Eng 391:114570

Agushaka JO, Ezugwu AE, Abualigah L (2023) Gazelle optimization algorithm: a novel nature-inspired metaheuristic optimizer. Neural Comput Appl 35(5):4099–4131

Khan EA, Shambour MK (2023) An optimized solution for the transportation scheduling of pilgrims in Hajj using harmony search algorithm. J Eng Res 11(2):100038

Mahendhiran PD, Kannimuthu S (2018) Deep learning techniques for polarity classification in multimodal sentiment analysis. Int J Inf Technol Dec Mak 17(03):883–910

Arunkumar PM, Chandramathi S, Kannimuthu S (2019) Sentiment analysis-based framework for assessing internet telemedicine videos. Int J Data Anal Tech Strateg 11(4):328–336

Naeem S et al (2021) Machine learning-based USD/PKR exchange rate forecasting using sentiment analysis of Twitter data. Comput Mater Contin 67(3):3451–3461

Mahendhiran PD, Subramanian K (2022) CLSA-CapsNet: dependency based concept level sentiment analysis for text. J Intell Fuzzy Syst 43(1):107–123

Download references

Acknowledgements

Not applicable

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Author information

Authors and affiliations.

Department of Computer Science and Artificial Intelligence, College of Computing, Umm Al-Qura University, 21955, Makkah, Saudi Arabia

Aisha M. Alasmari

College of Computing, Umm Al-Qura University, 21955, Makkah, Saudi Arabia

Norah S. Farooqi

Dar Al-Hekma University, Jeddah, Saudi Arabia

Department of Software Engineering, College of Computing, Umm Al-Qura University, 21955, Makkah, Saudi Arabia

Youseef A. Alotaibi

You can also search for this author in PubMed Google Scholar

Contributions

Aisha M. Alasmari: idea, searching, writing the manuscript. Norah S. Farooqi: idea, review the manuscript. Youseef Alotaibi; idea, review and modification the manuscript.

Corresponding author

Correspondence to Aisha M. Alasmari .

Ethics declarations

Conflict of interest.

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this article.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Alasmari, A.M., Farooqi, N.S. & Alotaibi, Y.A. Recent trends in crowd management using deep learning techniques: a systematic literature review. J. Umm Al-Qura Univ. Eng.Archit. (2024). https://doi.org/10.1007/s43995-024-00071-3

Download citation

Received : 07 December 2023

Accepted : 10 June 2024

Published : 20 June 2024

DOI : https://doi.org/10.1007/s43995-024-00071-3

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Crowed management
Crowd analysis
Crowd behaviors analysis
Social media
Find a journal
Publish with us
Track your research

IEEE Account

Change Username/Password
Update Address

Purchase Details

Payment Options
Order History
View Purchased Documents

Profile Information

Communications Preferences
Profession and Education
Technical Interests
US & Canada: +1 800 678 4333
Worldwide: +1 732 981 0060
Contact & Support
About IEEE Xplore
Accessibility
Terms of Use
Nondiscrimination Policy
Privacy & Opting Out of Cookies

A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity. © Copyright 2024 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to upgrade your browser .

We're Hiring!
Help Center

Crowd Management

Most Cited Papers
Most Downloaded Papers
Newest Papers
Scada Security Follow Following
eGovernance Follow Following
Disaster risk reduction Follow Following
Flood Risk Management Follow Following
Event Safety Follow Following
Licensing Act 2003 Follow Following
Crowd Behaviour and Psychology Follow Following
Crowd Psychology Follow Following
Crowd Safety Follow Following
Crowd behavior Follow Following

Enter the email address you signed up with and we'll email you a reset link.

Academia.edu Journals
We're Hiring!
Help Center
Find new research papers in:
Health Sciences
Earth Sciences
Cognitive Science
Mathematics
Computer Science
Academia ©2024

IMAGES

Crowd Management in Sport Facilities Free Essay Example
UAS Crowd Management Course 06/12/23
What is Crowd Managment? Crowd Management Meaning
What is Crowd Simulation? Crowd Simulation Meaning
Crowd management- In Events- Strategies, Reasons, Benefits, Why
Role of Crowd Sourcing Essay Sample

VIDEO

Management Techniques for Crowd Work
Crowd Management Training Module 1
CROWD CONTROL MANAGEMENT: HANDLING BIG CROWD
Crowd Safety
Learn more about BEST Crowd Management
People and Crowd Management Planning

COMMENTS

What Is Crowd Management?
For the purposes of this book, crowd management is defined in the following way: Proactive security activities that bring safety and comfort to individuals by facilitating efficient movement of crowds. Various characteristics of crowds are discussed in the following chapters, and the term is used in the broadest possible sense of the word.
Introduction to Crowd Management
Fundamental aspects related to crowd management are presented using simple concepts requiring little or no knowledge of mathematics or engineering. ... He has published several journal articles and conference papers, and has one patent registered in his name. In 2021, he had been co-awarded the Ig Nobel Prize in kinetics for a research on the ...
PDF Crowd Management
monitor behavior and disposition, and ensure that they are aware of any changes in crowd behavior or intent. 8. Audio and video recording of agency crowd response should be considered for evidentiary purposes. 9. Mass arrests shall be avoided, unless necessary. 10. Officers shall ensure that a means of egress for all individuals is present at ...
(PDF) Crowd Management -Navigating Challenges and Implementing Best
Crowd management refers to the systematic planning, organizati on, and implementation of measures to. control and regulate the movement, behavior, and activities of a large group of people in a ...
PDF What Is Crowd Management?
real-time crowd management but also for off-site crowd management and advance planning. In this book, we try to cover all these aspects. The concepts that we have touched upon here (including preventive ﬂow design to identify and eliminate risks) will be described in more detail in the following chapters. 1.4 Stakeholders and Crowd Information
Place crowd safety, crowd science? Case studies and application
Furthermore, crowd management deals with the strategic, tactical and operational handling of crowds ensuring safety in an uncompromising yet efficient manner. Crowd risk analysis is considered to be an important aspect of crowd monitoring and management (Smith, 2003). A number of studies highlight the shortcomings of the traditional approach ...
Crowd Management: Understanding Attitudes and Behaviors
ISSN: 1936-1610 print / 1936-1629 online. DOI: 10.1080/19361610.2014.913229. Crowd Management: Understanding. Attitudes and Behaviors. BRIAN F. KINGSHOTT. Grand Valley State University, Grand ...
A landscape of crowd-management support: An integrative approach
The papers included in this review aim to contribute to crowd management and together represent the modeling diversity in crowd research. The extended set of papers were extracted from the papers found in the review paper references (310) and in the Safety Science journal (144).
Crowd Management
Crowd management is an interdisciplinary discipline that requires a grasp of technical and technological issues since crowd behavior and flow management entail both psychological and social elements. A wide range of crowd management best practices and challenges for effective systems are covered in this article.
A landscape of crowd-management support: An integrative approach
papers focusing on only one or, at the most, two aspects of crowd management. Bellomo and Dogbe [2011]; Duives et al. [2013]; Challenger et al. [2009b]; V enuti and Bruno [2009] and Alsnih and Sto-
Crowd control
Crowd control is a public security practice in which large crowds are managed in order to prevent the outbreak of crowd crushes, affray, fights involving drunk and disorderly people or riots. Crowd crushes in particular can cause many hundreds of fatalities. [ 1] Effective crowd management is about managing expected and unexpected crowd ...
Crowd Management
The articles in this special section focus on crowd management in areas that are increasingly experiencing growths in populations. In its 2018 Revision of the World Urbanization Prospects, the United Nations projects that 68 percent of the world's population is expected to be living in urban spaces by 2050, compared to the current 54 percent and 30 percent in 1950. The main drive for ...
What is Crowd Management and Why is It Important?
What is Crowd Management? "Crowd control" is the prevailing term to define a major responsibility of facility management in stadiums and arenas. However, managing and assisting crowds—a practice known as crowd management—is a more proactive approach and often more effective than trying to control a crowd after things go wrong.
The Dangerous Power of Crowds
John Seabrook on crowd studies, human crushes, and the effort to prevent stampedes and deaths in places where large groups gather, such as stadiums and festival grounds.
(DOC) Crowd Science and risk management
Crowd Science and risk management. One of the key elements of crowd modelling is to understand the capacity of the space, how quickly it will fill and what time it will take to reach critical density1. This essay will examine, discuss and make recommendations on a specific venue site from the point of view of a consultant, examining the crowd ...
Recent trends in crowd management using deep learning ...
Crowd management has become an integral part of urban planning in abnormality in the crowd and predict its future issues. Big data in social media is a rich source for researchers in crowd data analysis. In this systematic literature review (SLR), modern societies. It can organize the flow of the crowd, perform counting, recognize the related works are analyzed, which includes crowd management ...
Crowd Management: A New Challenge for Urban Big Data Analytics
The increasing availability of tremendous amounts of data generated by people, vehicles, and things have provided unprecedented opportunities for understanding human behavior in the urban environment. At the same time, crowd management systems can benefit city planning, emergency control, and mobile network design. In this work, we exploit urban data as a way of analyzing crowd behavior. We ...
(PDF) Crowd safety and crowd risk analysis
The Crowd is a feature of humankind's history because humans are naturally gregarious entities who continually seek to be socialized (Templeton et al., 2018;Raineri, 2022). Its awareness and ...
Crowd Control At Events
Essay Writing Service. Poor crowd management can lead to crushing, injury and possible death. Events such as the Hillsborough Disaster (1989) resulted in the death of 96 football fans and hundreds more being injured. Lord Chief Justice Taylor identified the failure of police crowd control as a major factor in the tragedy (BBC 2009).
Recent trends in crowd analysis: A review
Group behavior analysis. Group behavior analysis is a subtopic of crowd behavior analysis. We highlight the recent trends in group behavior analysis through the review of Borja-Borja, Saval-Calvo, and Azorin-Lopez (2017). Borja et al. associate the nature of a human action to the number of individuals that perform it.
Digital Scholarship @UNLV
Digital Scholarship @UNLV | UNLV Libraries
Crowd Management Research Papers
The aim of the paper is to facilitate a theoretical approach of crowd management as research topic in social geography and spatial planning. The planning and realization of mass events are challenging, especially in regard to safety and security issues.
Understanding crowdsourcing projects: A review on the key design
The crowd refers to the people that participate in a crowdsourcing activity. The crowd is one of the most important actors in the crowdsourcing system (Zhao & Zhu, 2014a). The success of any crowdsourcing initiative largely depends on the ability to attract and motivate a crowd to develop solutions (Ford et al., 2015).

Place crowd safety, crowd science? Case studies and application

Design/methodology/approach

Practical implications

Originality/value

1. Introduction

2. Literature review

2.2 Crowd risk

2.3 Crowd science

2.3.1 Crowd modelling.

2.3.2 Crowd counting.

2.3.3 Crowd monitoring and management (including crowd risk analysis).

2.4 The limits of current knowledge in the field

3. Research methodology

4. Findings

4.1 Case 1 – Sydney Olympics project (2000)

4.2 Case 2 – Canary Wharf project (2003)

4.3 Case 3 – Murrayfield stadium project (2018)

5. Discussion and final remarks

5.1 What do we learn from the application of the design, information, management-ingress, circulation, egress model?

5.2 What works, in what contexts?

5.3 What does not work?

5.4 What limitations this model presents?

Further readings

Corresponding author

Report an issue or find answers to frequently asked questions

Crowd Management – Navigating Challenges and Implementing Best Practices in Crowd Management

Namreen Asif

Namreen Asif (Contact Author)

Do you have a job opening that you would like to promote on SSRN?

Crush Point

Recent trends in crowd management using deep learning techniques: a systematic literature review

Cite this article

Similar content being viewed by others

A Review of Deep Learning Techniques for Crowd Behavior Analysis

Video analytics using deep learning for crowd analysis: a review

1 Introduction

2 Research methodology

2.1 Phase (1): preliminary search

2.2 Phase (2): investigative search

2.3 Phase (3): study quality assessment

2.4 Phase (4): analysis of search

3 Related work

4 A comprehensive study of crowd management

4.1 Crowd detection

4.2 Crowd statistics

4.2.1 Crowd counting

4.2.2 Crowd density estimation

4.3 Crowd scene analysis

4.3.1 Crowd monitoring and tracking

4.3.2 Crowd behavior analysis

4.3.3 Crowd abnormality detection

4.3.4 Group activity detection

4.4 Social media-based analysis

4.4.1 Opinion/sentiment analysis

4.4.2 Geo-located analysis

4.5 Crowd sound emotion recognition

4.6 Crowd management at Hajj event

5 Results analysis of comprehensive study of crowd management

5.1 Crowd statistics

5.2 Crowd scene analysis

5.3 Social media-based analysis

5.4 Crowd sound emotion recognition

5.5 Crowd management at Hajj event

5.6 Addressing of used methodologies limitations

6 Discussion

6.1 Current study vision discussion

7 Conclusion

Availability of data and materials

Acknowledgements

Author information

Contributions

Corresponding author

Ethics declarations

Additional information

Rights and permissions

About this article

Share this article

IEEE Account

Purchase Details

Profile Information