Désormais, l'analyse des scènes denses s'impose incontestablement comme une tâche importante pour contrôler et gérer les foules. Notre recherche a pour objectifs d'apporter des solutions à l'estimation de la densité de la foule et de prouver l'utilité de cette estimation comme préalable pour d'autres applications. Concernant le premier objectif, afin de cerner les difficultés de la détection de personnes dans une foule, on se focalise sur l'estimation de la densité basée sur un niveau d'analyse bas. Dans un premier temps, on démontre que nos approches sont plus adéquates que les méthodes de l’état de l’art que ce soit pour compter les individus ou pour estimer le niveau de la foule. Dans un second temps, nous proposons une approche innovante dans laquelle une estimation locale au niveau des pixels remplace l'estimation au niveau global de la foule ou le nombre des personnes. Elle est basée sur l’utilisation des suivis de caractéristiques visuelles dans une fonction de densité. Notre recherche a également pour objectif d'utiliser la densité comme information supplémentaire pour affiner d'autres tâches. D'abord, nous avons utilisé la mesure de la densité qui comporte une description pertinente à la répartition spatiale des individus afin d'améliorer leur détection et leur suivi dans les foules. Ensuite, en prenant en compte la notion de la protection de la vie privée, nous ajustons le niveau de floutage en fonction de la densité de la foule. Enfin, nous nous appuyons sur l’estimation locale de la densité ainsi que sur le mouvement en tant qu'attributs pour des applications de haut niveau telles que la détection des évolutions et la reconnaissance des événements. / Crowd analysis has recently emerged as an increasingly important problem for crowd monitoring and management in the visual surveillance community. In this thesis, our objectives are to address the problems of crowd density estimation and to investigate the usefulness of such estimation as additional information to other applications. Towards the first goal, we focus on the problems related to the estimation of the crowd density using low level features in order to avert typical problems in detection of high density crowd. We demonstrate in this dissertation, that the proposed approaches perform better than the baseline methods, either for counting people, or alternatively for estimating the crowd level. Afterwards, we propose a novel approach, in which local information at the pixel level substitutes the overall crowd level or person count. It is based on modeling time-varying dynamics of the crowd density using sparse feature tracks as observations of a probabilistic density function. The second goal is to use crowd density as additional information to complement other tasks related to video surveillance in crowds. First, we use the proposed crowd density measure which conveys rich information about the local distributions of persons to improve human detection and tracking in videos of high density crowds. Second, we investigate the concept of crowd context-aware privacy protection by adjusting the obfuscation level according to the crowd density. Finally, we employ additional information about the local density together with regular motion patterns as crowd attributes for high level applications such as crowd change detection and event recognition.

This thesis investigates variation in human cooperative behaviour in naturally occurring contexts. I critically assess the prevailing consensus on human cooperation derived from laboratory games (such as the dictator and public goods games), by identifying real life analogues and conducting extensive field observation and experiments. My second chapter investigates the importance of context on social behaviour by taking a commonly used laboratory game, the dictator game, and studying analogous behaviour, giving to mendicants in the street. I conclude that individuals cooperate less in the wild than they do in the laboratory and that monetary pay-offs are important in cooperative decision-making. My third chapter examines how social cues influence peoples' likelihood of giving to mendicants. I conclude that increased group size and crowd density negatively affect donation behaviour. My fourth chapter investigates dog fouling in public parks to understand the causes of variation in cheating in a naturally occurring public goods game. I conclude that despite evidence that a social game is being played, the cues that influences decisions are unclear, and behaviour may depend on local social norms. My fifth chapter investigates social influences on red light jumping by cyclists at pedestrian crossings. I find that the probability of cheating is higher with fewer observers and when other cyclists also cheat.

This study investigates how people's perceived sense of safety affects their use of and feeling of safety when they are in a park. The chosen location for the case study research is Karlstad City Garden. Currently, Karlstad City Garden is underutilized and was described as unsafe in a survey conducted by the municipality of Karlstad in 2018 (Karlstad kommun 2019). The purpose of this study is to examine visitors' perceived feelings of safety when they are in Karlstad City Garden. The study will focus on four factors; lighting, lack of people, maintenance, and visibility to provide what effect they have on people when they visit a public place. Previous research, including the municipality's safety survey, identified these four factors as significant contributors to visitors' feelings of safety. This study aims to gain a better understanding of why visitors perceive themselves as safe or unsafe in the park and how these four factors influence their perceived sense of safety. To investigate visitors' experiences of the park, a questionnaire was distributed through social media, and QR codes were distributed in the park to reach a diverse range of respondents. Additionally, observations were conducted as part of the study, primarily to complement the questionnaire and strengthen the findings. The study findings revealed that the time of day and the four factors mentioned above influenced visitors' perceived sense of safety in the park. During the day, more people were present, which made visitors feel safer. In the evening, the park was largely empty, and there were few visitors. The feelings of safety during this time could be attributed to the low population in the park, inadequate lighting in most areas, and reduced visibility during the evening. The poor visibility and lighting could be a result of insufficient maintenance, particularly in the densely vegetated southern part of the park. As a result, many people may have avoided visiting the park in the evening due to feeling unsafe.

[ES] Esta tesis presenta un nuevo modelo llamado "Modelo dinámico de interacción social y multitud urbana (DUCSIM)", que tiene como objetivo calcular la densidad de multitudes y descifrar las redes sociales en entornos oportunistas. Con la creciente similitud de los dispositivos electrónicos conectados a Internet y la influencia generalizada de las redes sociales en línea, se ha creado un enorme rastro digital. Las huellas digitales basadas en la movilidad humana y el mayor uso de sistemas de comunicación inalámbrica como 3G, 4G y 5G forman una rica base de datos que puede analiarse. Estas huellas digitales ofrecen una forma única de modelar los patrones de multitud dentro de diferentes contextos, como asambleas espontáneas en espacios públicos y escenarios planificados, como en el caso de los megaeventos. El estudio se centra en el desafío de las reuniones multitudinarias oportunistas, donde las personas se congregan por diferentes motivos sin planificación; manifiestan sus movimientos de forma dinámica e inesperada. El análisis del comportamiento humano en las ciudades modernas y desarrolladas requiere que estas reuniones se produzcan en centros comerciales, cruces de carreteras y flash mobs. El análisis macroscópico de la densidad de multitudes basado en datos de las torres de telefonía móvil sirve como primera etapa para delinear el marco DUCSIM. Se adopta el método Median-of-Median (M-o-M) para mayor solidez, ya que este análisis implica umbrales de conteo bruto de multitudes diario y semanal. Las densidades de multitud se clasifican en cuartiles para mostrar distintos grados de distribución de la multitud. A través del análisis macroscópico, el marco avanza hacia el análisis de movilidad acumulativa de multitudes. La dinámica del movimiento de multitudes se mide cambiando las señales de las torres de telefonía movil y formulando un mapa de densidad de multitudes para pronosticar sus movimientos posteriores. Examina el microanálisis del movimiento individual y las relaciones interpersonales a menor escala. Incluye asignar personas a torres de telefonía móvil y formar gráficos de interacción social que infieren y actualizan las relaciones sociales. La parte más importante de DUCSIM radica en su capacidad de aprender y adaptarse dinámicamente para crear un modelo de representación novedoso que se adapte al patrón recién detectado. Esta flexibilidad ayuda a garantizar la relevancia del marco, que debe actualizarse continuamente. El modelado predictivo personalizado se combina con datos históricos que engloban la tesis. El marco utiliza densidades de multitudes anteriores y datos de movimiento para descubrir tendencias y predecir dinámicas de multitudes futuras, mejorando así la eficiencia de la planificación urbana, la respuesta a emergencias o las ciudades inteligentes. El marco DUCSIM proporciona un método integral, flexible y de previsión para comprender y controlar los fenómenos de aglomeración urbana. Una forma moderna de análisis de datos que involucra varias fuentes de datos, respaldada por matemáticas rigurosas, hace que este método sea único para los estudios urbanos. Además, da impulso al ámbito académico y proporciona recomendaciones prácticas sobre la aplicación de esta metodología en la gestión y planificación de las ciudades modernas. / [CA] Aquesta tesi presenta un nou model anomenat "Dynamic Urban Crowd and Social Interaction Model (DUCSIM)", que té com a objectiu calcular la densitat de multituds i desxifrar xarxes socials en entorns oportunistes. Amb la creixent comú d'aparells electrònics enllaçats a Internet i la influència generalitzada de les xarxes socials en línia, s'ha creat un enorme rastre digital. Les traces digitals basades en la mobilitat humana i l'augment de l'ús de sistemes de comunicació sense fils com 3G, 4G i 5G formen una base de dades rica per ser analitzada. Aquestes traces digitals ofereixen una manera única de modelar els patrons de multituds en diferents contextos, com ara assemblees espontànies en espais públics i escenaris planificats, com en el cas dels megaesdeveniments. L'estudi se centra en el repte de les reunions multitudinàries oportunistes, on la gent es congrega per diferents motius sense planificació; manifesten els seus moviments de manera dinàmica i inesperada. L'anàlisi del comportament humà a les ciutats modernes i desenvolupades requereix que aquestes reunions es produeixin en centres comercials, cruïlles de carreteres i flash mobs. L'anàlisi macroscòpic de la densitat de multituds basada en dades de les torres de telefonía mòbil serveix com a primera etapa per descriure el marc DUCSIM. El mètode M-o-M s'adopta per a la robustesa, ja que aquesta anàlisi implica umbrals de recompte de multituds diaris i setmanals. Les densitats de multitud es classifiquen en quartils per mostrar diferents graus de distribució de multitud. Mitjançant l'anàlisi macroscòpic, el marc avança cap a l'anàlisi de la mobilitat acumulat de multituds. La dinàmica del moviment de la multitud es mesura canviant els senyals de les torres de telefonía mòbil i formulant un mapa de densitat de la multitud per preveure els seus moviments posteriors. Examina el microanàlisi del moviment individual i les relacions interpersonals a menor escala. Inclou assignar persones a torres de telefonía mòbil i formar gràfics d'interacció social que dedueixin i actualitzin les relacions socials. La part més important de DUCSIM està en la seua capacitat per aprendre i adaptar-se de manera dinàmica per crear un model de representació nou que s'adapte al patró recentment detectat. Aquesta flexibilitat ajuda a garantir la rellevància del marc, que s'ha d'actualitzar contínuament. El modelatge predictiu personalitzat es combina amb les dades històriques que engloben la tesi. El marc utilitza dades de moviment i densitats de multitud anteriors per descobrir tendències i predir les properes dinàmiques de multituds, millorant així l'eficiència de la planificació urbana, la resposta d'emergència o les ciutats intel·ligents. El marc DUCSIM proporciona un mètode complet, flexible i de previsió per entendre i controlar els fenòmens d'aglomeracions urbanes. Una forma moderna d'anàlisi de dades que inclou diverses fonts de dades, amb el suport de matemàtiques rigoroses, fa que aquest mètode sigui únic per als estudis urbans. A més, dóna un impuls a l'àmbit acadèmic i ofereix recomanacions pràctiques sobre l'aplicació d'aquesta metodologia en la gestió i planificació de la ciutat moderna. / [EN] This thesis presents a new framework called the "Dynamic Urban Crowd and Social Interaction Model (DUCSIM)," which is aimed at calculating crowd density and deciphering social networks in opportunistic environments. With the growing commonality of internet-linked electronic gadgets and the widespread influence of online social networks, an enormous digital trail has been created. The digital traces based on human mobility and the increased usage of wireless communication systems such as 3G, 4G, and 5G form a rich database to be analyzed. These digital traces offer a unique way of modelling the crowd patterns within different contexts, like spontaneous assemblies in public spaces and planned scenarios, as in the case of mega-events. The study focuses on the challenge of opportunistic crowd gatherings, where people congregate for different reasons without planning; they manifest their motions dynamically and unexpectedly. The analysis of human behaviour in modern, developed cities requires that these gatherings occur in malls, road junctions, and flash mobs. Macroscopic crowd density analysis based on data from MOBILE towers serves as the first stage in outlining the DUCSIM framework. The Median-of-Median (M-o-M) method is adopted for robustness as this analysis involves daily and weekly raw crowd count thresholds. Crowd densities are ranked in quartiles to show varying degrees of crowd distribution. Through the macroscopic analysis, the framework progresses to cumulative crowd mobility analysis. Crowd movement dynamics are measured by changing signals from MOBILE towers and formulating a crowd's density map to forecast its subsequent motions. It examines the micro-analysis of individual movement and interpersonal relations on a smaller scale. It includes assigning people to MOBILE towers and forming social interaction graphs that infer and update social relationships. The most important part of DUCSIM lies in its ability to dynamically learn and adapt to create a novel representation model to suit the newly detected pattern. This flexibility helps to ensure the relevancy of the framework, which must be continually updated. Custom predictive modelling combines with historical data that encompasses the thesis. The framework uses previous crowd densities and movement data to discover trends and predict upcoming crowd dynamics, thus improving urban planning efficiency, emergency response, or smart cities. The DUCSIM framework provides a comprehensive, flexible and forecasting method of understanding and controlling urban crowd phenomena. A modern form of data analysis involving several data sources, supported by rigorous mathematics, makes this method unique for urban studies. Moreover, it gives impetus to the academic sphere and provides practical recommendations concerning the application of this methodology within modern city management and planning. / Addepalli, L. (2024). An Architecture for Crowd Density Estimation in Heterogenous Opportunistic Environment [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/204747