Pedestrians In Crowd Research Articles

Towards more reliable evaluation in pedestrian detection by rethinking “ignore regions”

It remains a challenging task to detect pedestrians in crowds and it needs more efforts to understand why the detectors fail. When we perform an error analysis based on the traditional evaluation strategy, we find that it produces many misleading false positives, which in fact cover occluded pedestrians. The reason for this is that we usually have two kinds of annotations in the dataset: regular pedestrians (detection targets) labeled by full-body boxes and ignored pedestrians (NOT detection targets) labeled by visible boxes. Ignored pedestrians are labeled as an additional category termed the “ignore region”. Nevertheless, our detectors always predict a full-body box for each pedestrian. This gap results in the following case: when a detector successfully predicts a full-body box for those ignored pedestrians, a false positive is triggered due to the low overlap between the predicted full-body box and the labeled visible box for the ignored pedestrian. This becomes even more harmful as the detector improves and becomes more capable of locating occluded pedestrians. To alleviate this issue, we devise a new pedestrian detection pipeline, which considers the additional visible box at both the detection and evaluation stages. During detection, we predict an extra visible box apart from the full-body box for every instance; during evaluation, we employ visible boxes instead of full-body boxes to match the “ignore region”. We apply the new pipeline to dozens of detection methods and validate the effectiveness of our pipeline in reducing the over-reporting of false positives and providing more reliable evaluation results.

Pedestrian Detection by Exemplar-Guided Contrastive Learning.

Typical methods for pedestrian detection focus on either tackling mutual occlusions between crowded pedestrians, or dealing with the various scales of pedestrians. Detecting pedestrians with substantial appearance diversities such as different pedestrian silhouettes, different viewpoints or different dressing, remains a crucial challenge. Instead of learning each of these diverse pedestrian appearance features individually as most existing methods do, we propose to perform contrastive learning to guide the feature learning in such a way that the semantic distance between pedestrians with different appearances in the learned feature space is minimized to eliminate the appearance diversities, whilst the distance between pedestrians and background is maximized. To facilitate the efficiency and effectiveness of contrastive learning, we construct an exemplar dictionary with representative pedestrian appearances as prior knowledge to construct effective contrastive training pairs and thus guide contrastive learning. Besides, the constructed exemplar dictionary is further leveraged to evaluate the quality of pedestrian proposals during inference by measuring the semantic distance between the proposal and the exemplar dictionary. Extensive experiments on both daytime and nighttime pedestrian detection validate the effectiveness of the proposed method.

Dynamic Guidance Strategy for Pedestrian Travel in Large-Scale Activity under Harsh Environment

Large-scale activities such as the Winter Olympics are usually held in areas with low temperature or other harsh environments, which greatly affects the spectating experience of pedestrians. In order to improve the travel efficiency and reduce the safety risk of pedestrians, an adaptive information-distribution strategy of VMS (variable message sign) for road networks is proposed to guide the pedestrians. In the proposed strategy, the dynamic feedback mechanism between the VMS information distribution and the state of crowded pedestrians is established, and the dynamic optimization model of the VMS information release layout is formulated. To evaluate the effectiveness of the strategy, a multiagent-based simulation method is proposed. Through numerical simulation, it is found that the guidance strategy can improve the movement efficiency by adjusting releasing duration of VMS information or improving the information obedience rate of pedestrians. In this paper, a large-scale competition area in the Xiaohaituo Mountain in Beijing was taken as an example to simulate the scenarios of ingress and egress with and without the strategy. The results show that the average walking time and the road congestion can be significantly reduced in the road network with the strategy, and the proportion of pedestrians with shorter travel time can be increased. Therefore, the research can provide theoretical foundation and data support for managers to guide passenger flows and improve the spectating experience.

OpenCV Based Automatic Detection of Pedestrian Crossing Platform Using Congestion Monitoring - A Survey

Road traffic congestion and pedestrian accidents are one of the major issues being faced worldwide. The reasons behind these accidents are mainly due to the risk when crossing or walking on road in urban and rural areas under heavy traffic. In order to avoid such circumstances a new idea is proposed- “Automatic uplifting of pedestrian crossing platform using traffic congestion monitoring”. The pedestrian and traffic congestion is constantly monitored using an IR (Infrared) sensor module. At the time when pedestrian density is more, the traffic signal turns red for the vehicles and let the pedestrian walk through uplifted crossing platform. The proposed system consists of motorized platform placed on the zebra crossing line which automatically uplifts when the infrared sensor senses and count the crowded pedestrians at the signal point. This method ensures safety in pedestrian crossing and will also not let drivers to break the rules, which may lead to accidents

Double Mask R-CNN for Pedestrian Detection in a Crowd

Aiming at the difficulty of feature extraction and the limitation of NMS (nonmaximum suppression) in crowded pedestrian detection, a new detection network named Double Mask R-CNN based on Mask R-CNN with FPN (Feature Pyramid Network) is proposed in this article. The algorithm has two improvements: firstly, we add a semantic segmentation branch on the FPN to strengthen the feature extraction of crowded pedestrians; secondly, we design a rule to estimate the pedestrian visibility of detected image according to the human keypoints information, and this rule can cover binary mask on the image whose pedestrian visibility is less than a certain threshold. Then we input the masked image into the network to locate occluded pedestrians. Experimental results on the CrowdHuman dataset show that the log-average miss rate (MR) of Double Mask R-CNN is 13, 12% lower than the best results of other mainstream networks. Similar improvements on WiderPerson dataset are also achieved by the Double Mask R-CNN.

Learning Pixel-Level and Instance-Level Context-Aware Features for Pedestrian Detection in Crowds

Pedestrian detection in crowded scenes is an intractable problem in computer vision, in which occlusion often presents a great challenge. In this paper, we propose a novel context-aware feature learning method for detecting pedestrians in crowds, with the purpose of making better use of context information for dealing with occlusion. Unlike most current pedestrian detectors that only extract context information from a single and fixed region, a new pixel-level context embedding module is developed to integrate multi-cue context into a deep CNN feature hierarchy, which enables access to the context of various regions by multi-branch convolution layers with different receptive fields. In addition, to utilize the distinctive visual characteristics formed by pedestrians that appear in groups and occlude each other, we propose a novel instance-level context prediction module which is actually implemented by a two-person detector, to improve the one-person detection performance. Applying with these strategies, we achieve an efficient and lightweight detector that can be trained in an end-to-end fashion. We evaluate the proposed approach on two popular pedestrian detection datasets, i.e., Caltech and CityPersons. The extensive experimental results demonstrate the effectiveness of the proposed method, especially under heavy occlusion cases.

Understanding Human Behaviors in Crowds by Imitating the Decision-Making Process

Crowd behavior understanding is crucial yet challenging across a wide range of applications, since crowd behavior is inherently determined by a sequential decision-making process based on various factors, such as the pedestrians' own destinations, interaction with nearby pedestrians and anticipation of upcoming events. In this paper, we propose a novel framework of Social-Aware Generative Adversarial Imitation Learning (SA-GAIL) to mimic the underlying decision-making process of pedestrians in crowds. Specifically, we infer the latent factors of human decision-making process in an unsupervised manner by extending the Generative Adversarial Imitation Learning framework to anticipate future paths of pedestrians. Different factors of human decision making are disentangled with mutual information maximization, with the process modeled by collision avoidance regularization and Social-Aware LSTMs. Experimental results demonstrate the potential of our framework in disentangling the latent decision-making factors of pedestrians and stronger abilities in predicting future trajectories.

Collective phenomena in crowds-Where pedestrian dynamics need social psychology.

This article is on collective phenomena in pedestrian dynamics during the assembling and dispersal of gatherings. To date pedestrian dynamics have been primarily studied in the natural and engineering sciences. Pedestrians are analyzed and modeled as driven particles revealing self-organizing phenomena and complex transport characteristics. However, pedestrians in crowds also behave as living beings according to stimulus-response mechanisms or act as human subjects on the basis of social norms, social identities or strategies. To show where pedestrian dynamics need social psychology in addition to the natural sciences we propose the application of three categories–phenomena, behavior and action. They permit a clear discrimination between situations in which minimal models from the natural sciences are appropriate and those in which sociological and psychological concepts are needed. To demonstrate the necessity of this framework, an experiment in which a large group of people (n = 270) enters a concert hall through two different spatial barrier structures is analyzed. These two structures correspond to everyday situations such as boarding trains and access to immigration desks. Methods from the natural and social sciences are applied. Firstly, physical measurements show the influence of the spatial structure on the dynamics of the entrance procedure. Density, waiting time and speed of progress show large variations. Secondly, a questionnaire study (n = 60) reveals how people perceive and evaluate these entrance situations. Markedly different expectations, social norms and strategies are associated with the two spatial structures. The results from the questionnaire study do not always conform to objective physical measures, indicating the limitations of models which are based on objective physical measures alone and which neglect subjective perspectives.

Influence of Long-range Interactions on Strategy Selection in Crowd

An order--disorder phase transition is observed for Ising-like systems even for arbitrarily chosen probabilities of spins flips [K. Malarz et al, Int. J. Mod. Phys. C 22, 719 (2011)]. For such athermal dynamics one must define $(z+1)$ spin flips probabilities $w(n)$, where $z$ is a number of the nearest-neighbours for given regular lattice and $n=0,\cdots,z$ indicates the number of nearest spins with the same value as the considered spin. Recently, such dynamics has been successfully applied for the simulation of a cooperative and competitive strategy selection by pedestrians in crowd [P. Gawro\'nski et al, Acta Phys. Pol. A 123, 522 (2013)]. For the triangular lattice ($z=6$) and flips probabilities dependence on a single control parameter $x$ chosen as $w(0)=1$, $w(1)=3x$, $w(2)=2x$, $w(3)=x$, $w(4)=x/2$, $w(5)=x/4$, $w(6)=x/6$ the ordered phase (where most of pedestrians adopt the same strategy) vanishes for $x>x_C\approx 0.429$. In order to introduce long-range interactions between pedestrians the bonds of triangular lattice are randomly rewired with the probability $p$. The amount of rewired bonds can be interpreted as the probability of communicating by mobile phones. The critical value of control parameter $x_C$ increases monotonically with the number of rewired links $M=pzN/2$ from $x_C(p=0)\approx 0.429$ to $x_C(p=1)\approx 0.81$.

Method of Bottleneck Identification and Evaluation During Crowd Evacuation Process

Accidents and natural disasters such as fire and earthquakes post more and more threats to the transportation infrastructures. Under such circumstance, the evacuation process for the crowded pedestrians in these infrastructures may be affected by the facility failure, resulting in partially damaged evacuation network, which at last forms bottlenecks and decreases the evacuation efficiency. Thus to improve the safety and comfortable level of crowd evacuation in transportation infrastructures, a new method of bottleneck identification and evaluation for evacuation network is proposed. The proposed method takes into account the facility failure induced crowd redistribution and explores the structure importance of the evacuation network component. The method of bottleneck identification is also applicable to other similar transportation networks.

Network Based Passenger Flow Modeling for Station Crowdedness Evaluation

The pedestrian comfort and safety level in mass transit stations is now attracting more and more attention in the past few years for the reason that the congested pattern in stations resulted from complex pedestrian activities including using the ticket add value machine, buying a ticket, shopping, waiting for services etc. Thus multi-directional pedestrian flows interact with each other and finally formed a relatively complex pedestrian distribution. To evaluate the crowdedness of the station for the convenient of crowd control, the crowded pedestrians are simulated with a heterogeneous network model in combined with a collision avoidance model. Based on the network model, a spatial configuration related station network is constructed, on which pedestrian flow can be formulized. At last, a station flow scenario is simulated and discussed with the formulized model to evaluate the complex activities induced crowdedness in metro station.

Network Based Passenger Flow Modeling for Station Crowdedness Evaluation

The pedestrian comfort and safety level in mass transit stations is now attracting more and more attention in the past few years for the reason that the congested pattern in stations resulted from complex pedestrian activities including using the ticket add value machine, buying a ticket, shopping, waiting for services etc. Thus multi-directional pedestrian flows interact with each other and finally formed a relatively complex pedestrian distribution. To evaluate the crowdedness of the station for the convenient of crowd control, the crowded pedestrians are simulated with a heterogeneous network model in combined with a collision avoidance model. Based on the network model, a spatial configuration related station network is constructed, on which pedestrian flow can be formulized. At last, a station flow scenario is simulated and discussed with the formulized model to evaluate the complex activities induced crowdedness in metro station.

Automatic Clustering Method of Abnormal Crowd Flow Pattern Detection

To ensure the safety of pedestrians in crowds under emergency situation is one of the most important concerns of researchers of building safety, traffic facility design as well as pedestrian traffic planning. Simulations as a consequence have been extensively studied to locate potential locations where danger might happen. However when abnormal pedestrian flow pattern occurs, control or guidance methods should be provided to other pedestrians based on the real-time information about the flow pattern. Thus in the present paper an automatically clustering method for detecting abnormal flow pattern is proposed. The method takes into account the interrelation between individual behavior and global pedestrian flow pattern. As a result it is indicated by the experimental and testing videos that the method is robust and efficient.

Psychological model for animating crowded pedestrians

This paper proposes a psychological model for simulating pedestrian behaviors in a crowded space. Our decision-making scheme controls plausible avoidance behavior depending on the positional relati...

Psychological model for animating crowded pedestrians

AbstractThis paper proposes a psychological model for simulating pedestrian behaviors in a crowded space. Our decision‐making scheme controls plausible avoidance behavior depending on the positional relations among surrounding persons, on the basis of a two‐stage personal space and a virtual memory structure as proposed in social psychology. Our system determines pedestrian walking speed with the crowd density to imitate the measured data in urban engineering, and automatically generates plausible motions of the individual pedestrian by composing a locomotion graph with motion capture data. Our approach based on psychology and a variety of actual measurements can increase the accuracy of simulation at both the micro and macro levels. Copyright © 2005 John Wiley & Sons, Ltd.

フィールド調査にもとづく通過しにくさの解析 : 群集流動横断時の歩行特性に関する研究 その2

The purpose of this paper is to clarify the level of walking difficulty through crowd flow. Four subjects crossed a crowd flow in a terminal station at four separate penetration angles each; 45 degrees, 90 degrees, 135 degrees. The crossing pedestrian's difficulty was analyzed from both the human movement and psychological aspects. The result of the analysis is as follows : 1) Mental evaluation of crossing difficulty has a close relation to the crowd density. The perception of crossing difficulty showed the greatest change between 0.25 person/m^2 and 1.0 person/m^2. 2) Mental evaluation of a 45 degree crossing angle was more favorable than one of 135 degrees. 3) The results of physical analysis were the same as mental evaluation. 4) The crossing subject tended to cross behind crowd pedestrians. In the case of high density, the avoidance behavior becomes complex, using changes of speed and locus.

Tracking Multiple Pedestrians in Crowd

This paper proposes a robust method for tracking many pedestrians viewed from an upper oblique angle. Persons are extracted while they are isolated, and they are tracked forward and backward. In tracking, if the region of a tracked pedestrian is ambiguous because of overlapping with other pedestrians, plural candidates for a pedestrian path are generated, and false candidates are eliminated in other frames. We make an experiment using real image sequences to show the effectiveness of the method.