Pedestrian Training Research Articles

Homomorphic federated learning schemes enabled pedestrian and vehicle detection system

Intelligent transport systems are increasingly being used in practice these days. Fog nodes and cloud servers collect real-time pedestrian and vehicle data and train them based on machine learning models. Existing pedestrian and vehicle detection systems need more security, less resource leakage, and faster processing. This paper proposes a homomorphic, secure, federated learning-enabled pedestrian detection system named HMFLS. The HMFLS consists of base stations (BS) and homogeneous (homo) federated learning servers with weights, surveillance, and traffic light components, and these nodes serve as data generation and processing sources. Homomorphic encryption is a cryptographic technique that allows computations on encrypted data without decryption. In other words, it enables computations on encrypted data while preserving the privacy and confidentiality of the information. The HMFLS exploits Generative Adversarial Networks to train pedestrian and vehicle images and extract features based on VGG19 from fog nodes and surveillance sensor images. We trained the different tasks and weights based on the model’s 28,000 pedestrian and vehicle images. The goal is to identify pedestrians and vehicles in the system. The interface is based on an Android-based application that can be easily integrated into different vehicles and mobile phones. Simulation results demonstrated that HMFL performed well as compared to existing schemes (e.g., PEL, TFL-CNN, FLAV) in terms of security accuracy by 98%, resource leakage by 50%, and processing time by 52% for vehicle and pedestrian detection in the system.

Camera-irrelevant Feature Invariance Learning for Unsupervised Person Re-identification

Unsupervised person re-identification (re-ID) aims to match pictures of target pedestrians from some other cameras without any labels. The existing methods usually adopt a cluster-based pipeline to generate false labels for training. However, because of the differences in camera configuration, taking pictures Angle, illumination intensity and pedestrian background, unlabeled pedestrian training samples captured by different cameras are hard to classify into a cluster. To solve this problem, the paper proposes a camera-irrelevant feature learning framework, which learns camera-irrelevant features and camera-irrelevant distances through network training and clustering. Specifically, during training, images of other camera styles are generated, which enriches cross-camera learning. During clustering, we propose a camera-irrelevant distance matrix, which reduces the inter-class variance caused by camera variance.

The Perspective Projects Promoting Sustainable Mobility by Active Travel to School on the Example of the Southern Poland Region

The paper presents a discussion concerning the development of projects regarding active travel with the policy of sustainable mobility, with active school transport in southern Poland being taken into special consideration. The implementation of the idea of sustainable school travel planning involves linking several social groups, including traffic planners and organizers, school administrators, governments, parents, and children. This, in turn, requires considering the criteria reported by all parties when choosing a scenario for the region’s development in terms of transport solutions. The following study was based on the methodological foundations of multicriteria decision-making analysis. The research purpose of this paper is to identify and classify the actions, policies, and scenarios of active travel projects for the sustainable development of mobility based on the cities of southern Poland. The evaluation was carried out through expert methods with multicriteria decision-making tools based on the MULTIPOL (MULTI-criteria and POLicy) prospective analysis technique. It allowed for the selection of the most probable policy, which covered the six actions considered by the experts to be the most significant for the development of the active school transport system in the analyzed region. Such actions are as follows: identification of walking school bus routes, modernization of crosswalk lines, planning of walking and cycling routes to schools, promoting safety educational programs, distribution of active school transport booklets and cycling, and pedestrian skills training workshops. The paper presents a new method to evaluate the policies and actions regarding promoting sustainable (active) travel to school. The innovative approach results from assembling a mixed group of people (stakeholders) as experts. The mix of experts consisting of users (pupils and parents), practitioners, and scientist experts in this field allowed us to score policies, actions, and scenarios, enabling a wider spectrum of assessment than before.

Swin Transformer Based on Two-Fold Loss and Background Adaptation Re-Ranking for Person Re-Identification

Person re-identification (Re-ID) aims to identify the same pedestrian from a surveillance video in various scenarios. Existing Re-ID models are biased to learn background appearances when there are many background variations in the pedestrian training set. Thus, pedestrians with the same identity will appear with different backgrounds, which interferes with the Re-ID performance. This paper proposes a swin transformer based on two-fold loss (TL-TransNet) to pay more attention to the semantic information of a pedestrian’s body and preserve valuable background information, thereby reducing the interference of corresponding background appearance. TL-TransNet is supervised by two types of losses (i.e., circle loss and instance loss) during the training phase. In the retrieval phase, DeepLabV3+ as a pedestrian background segmentation model is applied to generate body masks in terms of query and gallery set. The background removal results are generated according to the mask and are used to filter out interfering background information. Subsequently, a background adaptation re-ranking is designed to combine the original information with the background-removed information, which digs out more positive samples with large background deviation. Extensive experiments on two public person Re-ID datasets testify that the proposed method achieves competitive robustness performance in terms of the background variation problem.

Pedestrian Physical Education Training Over Visualization Tool

E-learning approaches are one of the most important learning platforms for the learner through electronic equipment. Such study techniques are useful for other groups of learners such as the crowd, pedestrian, sports, transports, communication, emergency services, management systems and education sectors. E-learning is still a challenging domain for researchers and developers to find new trends and advanced tools and methods. Many of them are currently working on this domain to fulfill the requirements of industry and the environment. In this paper, we proposed a method for pedestrian behavior mining of aerial data, using deep flow feature, graph mining technique, and convocational neural network. For input data, the state-of-the-art crowd activity University of Minnesota (UMN) dataset is adopted, which contains the aerial indoor and outdoor view of the pedestrian, for simplification of extra information and computational cost reduction the pre-processing is applied. Deep flow features are extracted to find more accurate information. Furthermore, to deal with repetition in features data and features mining the graph mining algorithm is applied, while Convolution Neural Network (CNN) is applied for pedestrian behavior mining. The proposed method shows 84.50% of mean accuracy and a 15.50% of error rate. Therefore, the achieved results show more accuracy as compared to state-of-the-art classification algorithms such as decision tree, artificial neural network (ANN).

APredictive Resource Allocation for Wireless Communications Systems.

Using event-driven computer simulations, we test a predictive resource allocation scheme for broadband wireless communications systems based on the Kalman Filter. We implemented the filter in six different scheduling algorithms and selected the best performing of them (EXP Rule and FLS) for a detailed analysis. We considered several simulation scenarios with velocities ranging from 0 to 350 km/h representing pedestrian, vehicular and high-speed train scenarios. We also considered different traffic profiles by changing the number of mobile devices between 10 and 250. Our simulations showed improvement in throughput, spectral efficiency, and a decrease in packet loss rates for two schedulers: the EXP Rule and the Frame Level Scheduler. Our predictor can be applied to any wireless system that performs centralized resource allocation based on data rate or any other quantity that needs to be estimated for resource distribution in a network.

Contrastive learning of graph encoder for accelerating pedestrian trajectory prediction training

In the area of pedestrian trajectory prediction, the hybrid structures of temporal feature extractor or spatial feature extractor have paved the way for the precise prediction model, and they are in larger and larger scale. Learning of specific feature encoding model not only influenced by the structure of the network, but also by the learning manners such as supervised learning and unsupervised learning. Previous works concentrated on more comprehensive encoders and more delicate designs of feature extractors. However, the mutual influence factors from the neighbour pedestrians associate with the distance to the centre pedestrian seldomly noticed. Most of the existed feature extractors in prediction models trained in the way of supervised learning other than unsupervised manners caused the problem that the extracted features are always handcrafted without the natural distinction of obscure situations. The graph contrastive accelerating encoder is proposed, which accelerates the pedestrian trajectory prediction training process of the state of the art method of spatio-temporal graph transformer networks. Employing the unsupervised contrastive learning process and the graph of neighbours representing distance affection of nearest and farthest pedestrian to the centre pedestrian, the graph contrastive accelerating encoder significantly shrinked the training time. Holding the final performance on to state of the art level, the proposed method let the lowest pedestrian trajectory prediction error show up in the obviously earlier training steps.

Vibration serviceability of a GFRP railway crossing due to pedestrians and train excitation

Glass Fibre-Reinforced Polymers (GFRPs) are a popular option for pedestrian bridges over railway lines as they cause little disruption for installation or maintenance. However, as they are typically lighter and less stiff than traditional materials there is concern about their dynamic response due to human induced actions and train buffeting. Due to a lack of experimental information, further data is needed if such bridges are to be used on future lines, especially with higher speed limits. This paper presents an experimental investigation of the response of a 14.5 m GFRP truss bridge due to pedestrian loading and train induced vibrations. Vibration modes of interest were identified from impact hammer tests. The vibration responses to a range of human loading scenarios, as well as multiple train passes, were measured. The vibration levels remained low under all conditions, demonstrating that this particular type of GFRP bridge is suitable for railway crossings and that further optimisations may be available to designers while meeting vibration serviceability limits. However, consideration of the loadings suggests GFRP bridges may be more susceptible to higher pedestrian harmonics than traditional structures, and that vibrations from train buffeting are likely to be a design consideration for future bridges over high speed lines.

Evaluation of pedestrian–train fatalities in the State of Maryland: A 5-year retrospective study

Few studies have been done on the incidences of train-related pedestrian fatalities throughout the United States, with no previous studies reported in the State of Maryland. A retrospective study was conducted at the statewide medical examiner's office in Maryland to evaluate the characteristics of train-related pedestrian fatalities from 2014 to 2018. The aim of the study was to analyze circumstances of deaths through the medicolegal death investigation and postmortem examination findings, including toxicological study, to help identify epidemiological characteristics of pedestrian–train-related fatalities. A total of 48 pedestrian-train deaths were identified. Of the 48 cases identified, 21 deaths (43.75%) were determined to be accident, 20 deaths (41.67%) were suicide, and 7 deaths (14.58%) whose manner of death could not be determined. Of the 21 accidental victims, 17 were male and 4 were female (M:F ratio = 4.3:1), 11 (52.38%) were white, 6 (28.57%) African American, and 4 Hispanic (19.05%), with age ranging from 16 to 58 years (mean age = 35). Of the 20 suicide victims, 17 were males and 3 females (M:F ratio = 5.7:1), 16 (80%) were white, 2 (10%) African American, and 2 (10%) Hispanic, with age ranging from 22 to 60 years (mean age = 40). The majority of accidents occurred during weekday evening rush hours between 4:00 pm and 7:00 pm, while the suicides showed no specific time frames. No specific peak for seasons of year was found in suicides or accidents. Postmortem toxicological studies showed that 52.38% (11/21 cases) of accidental victims were positive for alcohol and 30% (6/20 cases) of suicide victims were positive for alcohol. Manner of death could not be determined in 7 cases because of unclear circumstances of death. Thorough death scene investigation and complete postmortem examination, including comprehensive toxicological testing, is very important in all train-related pedestrian fatalities. Determining the manner of death can directly affect the outcome of civil ligation and dispersal of insurance benefits. The characteristic profiles of train-related pedestrian fatalities can also assist effective preventive measures on railway suicides and accidents.

Real-time pedestrian detection method based on improved YOLOv3

Pedestrian detection in image or video data is a very important and challenging task in security surveillance. The difficulty of this task is to locate and detect pedestrians of different scales in complex scenes accurately. To solve these problems, a deep neural network (RT-YOLOv3) is proposed to realize real-time pedestrian detection at different scales in security monitoring. RT-YOLOv3 improves the traditional YOLOv3 algorithm. Firstly, the deep residual network is added to extract vehicle features. Then six convolutional neural networks with different scales are designed and fused with the corresponding scale feature maps in the residual network to form the final feature pyramid to perform pedestrian detection tasks. This method can better characterize pedestrians. In order to further improve the accuracy and generalization ability of the model, a hybrid pedestrian data set training method is used to extract pedestrian data from VOC data set and train with INRIA pedestrian data set. Experiments show that the proposed RT-YOLOv3 method achieves 93.57% accuracy of mAP (mean average precision) and 46.52f/s (number of frames per second). In terms of accuracy, RT-YOLOv3 performs better than Fast R-CNN, Faster R-CNN, YOLO, SSD, YOLOv2 and YOLOv3. This method reduces the missed detection rate and false detection rate, improves the positioning accuracy, and meets the requirements of real-time detection of pedestrian objects.

How to Train Your Dragon: Best Practices in Pedestrian Classifier Training

The present year witnesses another milestone in Pedestrian detection's journey. It has achieved remarkable progress in the course of the past 15 years, and experts foresee an everyday use of numerous stemming applications within the next 15 years. Standing on the tipping point between yesterday and tomorrow pushes to the field's retrospect. Features, context, the combination of approaches, and feature data are responsible for most of the breakthroughs in pedestrian detection. If the first three elements cover the bulk of the literature, much fewer efforts have been dedicated to featuring data. In many aspects, the construction of a training set remains similar to what it was at the birth of the domain, and some related problems are not well studied and sometimes still tackled empirically. This paper gets down to the study of pedestrian classifier training conditions. More than a survey of existing training classifiers or features, our goal is to highlight impactful parameters, potential new research directions, and combination dilemmas. Our findings are experimentally verified on two major datasets: the INRIA and Caltech-USA datasets.

LCrowdV: Generating labeled videos for pedestrian detectors training and crowd behavior learning

We present a novel procedural framework to generate an arbitrary number of labeled crowd videos (LCrowdV). The resulting crowd video datasets are used to design accurate algorithms or training models for crowded scene understanding. Our overall approach is composed of two components: a procedural simulation framework to generate crowd movements and behaviors, and a procedural rendering framework to generate different videos or images. Each video or image is automatically labeled based on the environment, number of pedestrians, density, behavior (personality), flow, lighting conditions, viewpoint, type of noise, etc. Furthermore, we can increase the realism by combining synthetically-generated behaviors with real-world background videos. We demonstrate the benefits of LCrowdV over prior lableled crowd datasets by augmenting a real dataset with it and improving the accuracy in pedestrian detection and crowd classification. Furthermore, we evaluate the impact of removing the variety in different LCrowdV parameters to show the importance of the diversity of data generated from our framework. LCrowdV has been made available as an online resource.

A Pilot Randomized Controlled Trial Testing the Effectiveness of a Pedestrian Training Program That Teaches Children Where and How to Cross the Street Safely.

Pedestrian injury is a leading cause of injury-related mortality for children. This pilot randomized controlled trial tested the efficacy of a training program to teach where and how to cross safely. Using fully immersive virtual reality technology, 142 children 7-10 years of age were recruited, with 130 completing crossing measures before (pretest) and immediately after (posttest) training. Training comprised 1.5 hr, was tailored to each child's performance over trials, and focused on either where to cross (n = 44 children completed testing) or how to cross safely (n = 43); corresponding control groups comprised 22 and 21 children, respectively. Following training, children in the intervention groups completed additional tasks to test conceptual knowledge and generalization of learning. Children in the control groups spent the same time as those in training groups but played a video game that used the same game controller but provided no training in street crossing. The primary outcomes were errors in crossing at posttest, controlling for pretest error scores. Children in the intervention group made from 75% to 98% fewer errors at posttest than control children for all pedestrian safety variables related to where and how to cross safely, with effect sizes (incidence rate ratios) varying between 0.02 and 0.25. They also showed a generalization of what they had learned and applied this knowledge to novel posttraining situations. Training within a virtual pedestrian environment can successfully improve children's conceptual understanding and crossing behaviors for both where and how to cross streets safely.

Mistakes or deliberate violations? A study into the origins of rule breaking at pedestrian train crossings

Train pedestrian collisions are the most likely to result in severe injuries and fatalities when compared to other types of rail crossing accidents. However, there is currently scant research that has examined the origins of pedestrians’ rule breaking at level crossings. As a result, this study examined the origins of pedestrians’ rule breaking behaviour at crossings, with particular emphasis directed towards examining the factors associated with making errors versus deliberation violations. A total of 636 individuals volunteered to participate in the study and completed either an online or paper version of the questionnaire. Quantitative analysis of the data revealed that knowledge regarding crossing rules was high, although up to 18% of level crossing users were either unsure or did not know (in some circumstances) when it was legal to cross at a level crossing. Furthermore, 156 participants (24.52%) reported having intentionally violated the rules at level crossings and 3.46% (n=22) of the sample had previously made a mistake at a crossing. In regards to rule violators, males (particularly minors) were more likely to report breaking rules, and the most frequent occurrence was after the train had passed rather than before it arrives. Regression analysis revealed that males who frequently use pedestrian crossings and report higher sensation seeking traits are most likely to break the rules. This research provides evidence that pedestrians are more likely to deliberately violate rules (rather than make errors) at crossings and it illuminates high risk groups. This paper will further outline the study findings in regards to the development of countermeasures as well as provide direction for future research efforts in this area.

Learning a Part-Based Pedestrian Detector in a Virtual World

Detecting pedestrians with on-board vision systems is of paramount interest for assisting drivers to prevent vehicle-to-pedestrian accidents. The core of a pedestrian detector is its classification module, which aims at deciding if a given image window contains a pedestrian. Given the difficulty of this task, many classifiers have been proposed during the last 15 years. Among them, the so-called (deformable) part-based classifiers, including multiview modeling, are usually top ranked in accuracy. Training such classifiers is not trivial since a proper aspect clustering and spatial part alignment of the pedestrian training samples are crucial for obtaining an accurate classifier. In this paper, we first perform automatic aspect clustering and part alignment by using virtual-world pedestrians, i.e., human annotations are not required. Second, we use a mixture-of-parts approach that allows part sharing among different aspects. Third, these proposals are integrated in a learning framework, which also allows incorporating real-world training data to perform domain adaptation between virtual- and real-world cameras. Overall, the obtained results on four popular on-board data sets show that our proposal clearly outperforms the state-of-the-art deformable part-based detector known as latent support vector machine.

Making in‐class skills training more effective: The scope for interactive videos to complement the delivery of practical pedestrian training

AbstractSkills and awareness of young pedestrians can be improved with on‐street practical pedestrian training, often delivered in schools in the UK by local authorities with the intention of improving road safety. This training is often supplemented by in‐class paper‐based worksheet activities that are seen to be less effective than practical training in that they focus on knowledge acquisition rather than directly improving the correct application of safe pedestrian skills at the roadside. Previous research indicates that interactive video tools have the potential to develop procedural skills while offering an engaging road safety educational experience, which could positively impact on road crossing behaviour.In this paper, the design and development of a hazard‐identification interactive road safety training video targeting child road crossing skills is presented. The interactive video was shown to be an engaging training resource for 6‐ to 7‐year‐old children. The tool's scope for improving pedestrians' roadside skills is considered along with the wider implications for interactive video to aid safety training in other areas.

Children’s pedestrian route selection: Efficacy of a video and internet training protocol

ObjectiveThis randomized controlled trial examined one aspect of child pedestrian behavior, route selection across intersections, to evaluate whether a combination of widely-available videos and websites effectively train children in safe pedestrian route selection compared to active pedestrian safety control training and a no-contact control group. MethodsA sample of 231 seven- and eight-year-olds were randomly assigned to one of four groups: training with videos and internet websites, active control groups of individualized streetside training or training within a virtual pedestrian environment, or a no-contact control group. All training groups received six 30-min training sessions. Pedestrian route selection was assessed using two strategies, vignettes accompanied by illustrations and tabletop models of intersections, on three occasions: prior to intervention group assignment, immediately post-training, and six months after training. ResultsAlthough there were differences in route selection over time, no time by condition interaction effects were significant (ps>.05), suggesting children in the video/internet training group did not learn pedestrian route selection skills at a rate different from those in the other training groups or those in the no-contact control group. ConclusionSafe route selection is a critical component of pedestrian safety. Our results suggest children may not learn route selection from widely-available videos or websites. Explanations for the null finding and implications for both research and future practice are discussed.

Crossing a two-way street: comparison of young and old pedestrians

IntroductionChoosing a safe gap in which to cross a two-way street is a complex task and only few experiments have investigated age-specific difficulties. MethodA total of 18 young (age 19–35), 28 younger-old (age 62–71) and 38 older-old (age 72–85years) adults participated in a simulated street-crossing experiment in which vehicle approach speed and available time gaps were varied. The safe and controlled simulated environment allowed participants to perform a real walk across an experimental two-way street. The differences between the results for the two lanes are of particular interest to the study of visual exploration and crossing behaviors. ResultsThe results showed that old participants crossed more slowly, adopted smaller safety margins, and made more decisions that led to collisions than did young participants. These difficulties were found particularly when vehicles approached in the far lane, or rapidly. Whereas young participants considered the time gaps available in both lanes to decide whether to cross the street, old participants made their decisions mainly on the basis of the gap available in the near lane while neglecting the far lane. ConclusionsThe present results point to attentional deficits as well as physical limitations in older pedestrians. Several practical and have implications in terms of road design and pedestrian training are proposed.

The Development of Child Pedestrian Training in the United Kingdom 2002–2011: A National Survey of Local Authorities

Approximately two thirds of the children killed or seriously injured on Great Britain's roads are child pedestrians. Recognizing this issue, the Department for Transport introduced and evaluated a pilot child pedestrian training scheme, Kerbcraft, from 2002 to 2007. Kerbcraft aimed to teach roadside pedestrian skills to children between age 5 and 7 and was successfully trialled across 75 local authorities in England and Scotland. The UK government has since recommended Kerbcraft as the best practice example for roadside child pedestrian training. This article revisits a sample of local authorities that participated in the pilot and presents the findings from a new survey: identifying the status of child pedestrian training and assessing how the delivery, evaluation, and learning mechanisms have changed since the pilot. The results suggest that following on from the Kerbcraft scheme, the majority of local authorities continue to provide pedestrian training but in an adapted form—often delivering less training than that suggested by the Kerbcraft model. The implications are discussed.

Effects of Picture Prompts Delivered by a Video iPod on Pedestrian Navigation

Transportation access is a major contributor to independence, productivity, and societal inclusion for individuals with intellectual and developmental disabilities (IDD). This study examined the effects of pedestrian navigation training using picture prompts displayed through a video iPod on travel route completion with 4 adults with IDD. Results indicated a functional relation between picture prompts on the video iPod and pedestrian navigation skills to and from various locations. Maintenance data indicated all participants continued navigating trained routes for up to 232 days using the video iPod. Generalization measures to untrained routes were positive. Social validity data suggested iPod training was useful and practical for teaching independent pedestrian navigation skills. Last, limitations, suggestions for future research, and implications for practice were provided.