1,721,003 research outputs found
Gender Profiling of Pedestrian Dyads
No full textIn traffic safety community, behavioral differences between genders
have been attracting considerable attention in recent decades. Various empirical
studies have proven that gender has a significant relation to drivers’, cyclists’ or
pedestrians’ decision making, route choice, rule compliance, as well as risk taking/
perception. However, most studies examine behavior of individuals, and only
very few consider (pedestrian) groups with different gender profiles. Therefore,
this study investigates effect of gender composition of pedestrian dyads on the
tangible dynamics, which may potentially help in automatically understanding
and interpreting higher level behaviors such as decision making.We first propose
a set of variables to represent dyads’s physical/dynamical state. Observing empirical
distributions, we comment on the effect of gender interplay on locomotion
preferences. In order to verify our inferences quantitatively, we propose a gender
profile recognition algorithm. Removing one variable at a time, contribution of
each variable to recognition is evaluated. Our findings indicate that height related
variables have a more strict relation to gender, followed by group velocity and
inter-personal distance. Moreover, the “male” effect on dyad motion is found to
somehow diminish when the male is paired with a female
Social group behaviour of triads. Dependence on purpose and gender
Full text linkWe analysed a set of uninstructed pedestrian trajectories automatically tracked in a public area, and we
asked a human coder to assess their group relationships. For those pedestrians who belong to the groups, we asked
the coder to identify their apparent purpose of visit to the tracking area and apparent gender. We studied the quantitative
dependence of the group dynamics on such properties in the case of triads (three people groups) and compared
them to the two pedestrian group case (dyads), studied in a previous work. We found that the group velocity strongly
depends on relation and gender for both triads and dyads, while the influence of these properties on spatial structure
of groups is less clear in the triadic case. We discussed the relevance of these results to the modelling of pedestrian
and crowd dynamics, and examined the possibility of the future works on this subject
Effect of Tool Specificity on the Performance of DNN-Based Saliency Prediction Methods
No full textThis study focuses on the performance of saliency
models concerning a specific type of image, namely hand tools.
These objects are characterized by functionally distinct segments
with various manipulative roles (i.e. end-effectors) and parts
that are used to grasp and operate (i.e. handles). A highlighted
by various studies in behavioral science, en-effectors of tools
inherently draw humans’ attention. However, it remains unclear
whether saliency models effectively address this intrinsic
aspect. To shed light on this, we selected four recent notable
saliency models, i.e. EML-NET, SalGAN, DeepGaze IIE, and
DeepGaze III, known for their reliance on transfer learning. Our
aim is to evaluate their performance in capturing the influence
of semantic segments within tools. To conduct the assessment,
we carefully chose a set of images featuring tool and nontool
objects from a large standardized dataset and applied each
saliency model to these images. Subsequently, these images were
presented to a group of human participants, and empirical gaze
data was recorded. Finally, we evaluated the correspondence
(or discrepancy) between the saliency maps and the empirical
data using six different evaluation criteria (i.e. CC, NSS, LL,
IG, KL, SIM). Our findings reveal that, across all four models,
the accuracy in predicting saliency concerning tool images often
lags behind that of non-tool images. Moreover, two out of the
four models exhibited consistently lower accuracy in predicting
saliency on tool images compared to non-tool images across
all six evaluation criteria. This indicates a lack of adequate
consideration for the specificity of tools in these recent saliency
models, highlighting the necessity to propose solutions to rectify
this limitation
Gait synchronisation in pedestrian dyads: The influence of social interaction
Full text linkSpontaneous gait synchronisation is commonly observed in pedestrian groups and has been studied extensively in controlled settings. Here, we investigate this phenomenon in a natural environment using a pedestrian trajectory dataset collected with range sensors in a public space, along with annotations for social groups and their interaction levels. To quantify synchronisation, we analyze the lateral periodic swaying of pedestrians, computed as orthogonal displacements from smoothed trajectories. Using the Hilbert transform, we derive instantaneous phase of pedestrians' gait residuals and then compute relative phases for all dyads. Additionally, we calculate the Gait Synchronisation Index (GSI) to quantify the level of synchronisation between pedestrians. Results show significantly higher GSI, stronger phase locking around zero, and lower phase variance in dyads with high interaction levels compared to less interactive pairs and randomly chosen pairs of pedestrians. These findings highlight the role of social interaction in gait synchronisation and provide insights into crowd dynamics and motor coordination, with potential applications in socially-aware robotics and intelligent transportation systems
Estimating social relation from trajectories
Full text linkThis study focuses on social pedestrian groups in public spaces and makes an effort to identify the
social relation between the group members. We particularly consider dyads having coalitional or mating relation.
We derive several observables from individual and group trajectories, which are suggested to be distinctive for these
two sorts of relations and propose a recognition algorithm taking these observables as features and yielding an
estimation of social relation in a probabilistic manner at every sampling step. On the average, we detect coalitional
relation with 87% and mating relation with 81% accuracy. To the best of our knowledge, this is the first study to
infer social relation from joint (loco)motion patterns and we consider the detection rates to be a satisfactory
considering the inherent challenge of the problem
A computationally efficient approach for solving RBSC-based formulation of the subset selection problem
No full textThis study focuses on a specific type of subset selection
problem, which is constrained in terms of the rank bi-serial
correlation (RBSC) coefficient of the outputs. For solving such
problems, we propose an approach with several advantages such
as (i) providing a clear insight into the feasibility of the problem
with respect to the hyper-parameters, (ii) being non-iterative, (iii)
having a foreseeable running time, and (iv) with the potential
to yield non-deterministic (diverse) outputs. In particular, the
proposed approach is based on starting from a composition of
subsets with an extreme value of the RBSC coefficient (e.g. = 1)
and swapping certain elements of the subsets in order to adjust
into the desired range. The proposed method is superior to the
previously proposed RBSC-SubGen, which attempts to solve the
problem before confirming its feasibility, taking random steps,
and has unforeseeable running times and saturation issues
Developing a web application for RBSC-based solution of the subset selection problem
No full textIn this article, we introduce an application, which
implements the RBSC-SubGen algorithm on a web platform in
an easy-to-use manner. Originally, Furuya et al. proposed this
algorithm and demonstrated it on an sample scenario, where
a pair of vocabulary decks are constructed with a desired
difficulty relation out of a large corpus. In addition to such
applications, RBSC-SubGen can be used in a broad range of
applications. For instance, studies which require the recruitment
of a representative set of human subjects (e.g. drug testing,
consumer surveys) may benefit from this method in sampling
from the population. However, the deployment of the algorithm
in non-technical fields such as medicine or social science may be
difficult, since the publicly available algorithm implementation
target users skilled in software development. In that respect, with
the proposed web application the accessibility and disponibility
of the algorithm by users from non-technical fields are expected
to be facilitated considerably
Investigating the effect of various types of audio reinforcement on memory retention
No full textMost e-learning systems deliver solely visual information,
even though they boast a huge potential for supporting
the learners using various other capabilities (e.g. camera, speakers)
of the hosting platform (i.e. computer, smart phone etc.).
In this study, we focus deploying one such potential, namely
audio stimuli (informative and non-informative), for supporting
rote learning of different types of learning material (i.e. easy
verbal, hard verbal and numerical). Our results indicate that
audio stimuli do not provide a significant benefit for studying
easy verbal content, but there is a big room for improvement
concerning other content types (hard verbal and numerical).
Interestingly, despite the general implications of dual-coding
theory, human-readout of hard verbal contents is observed not
to provide any significant improvement over visual-only stimuli.
However, to our surprise, non-informative audio stimuli (i.e.
bell sound) are observed to provide an improvement, whereas
numerical content is observed to benefit in a similar way from
informative and non-informative audio. Based on these results,
in the future we aim developing an automatic learning support
system, which triggers the appropriate audio stimuli, taking in
consideration the type of content
Ecological data reveal imbalances in human–human collision avoidance due to dyads' social interaction
Full text linkHumans navigate public spaces safely and smoothly using complex collision avoidance strategies. Traditional models of human–human collision avoidance often draw from physics, relying on repulsive forces, but the effect of social factors on these strategies is not well understood. This study examines frontal encounters between single pedestrians and two-person groups (dyads), investigating the contributions of each party to collision avoidance and the impact of social interaction within the group. Using an ecological dataset of pedestrian trajectories, we measured deviations from a straight path as a proxy for collision avoidance. Our findings reveal a systematic imbalance and significant effects of social interaction on collision avoidance. Single pedestrians tend to prioritise trajectory efficiency in undisturbed situations and are the primary contributors to avoidance during encounters, adjusting their paths according to the dyad's interaction level. For dyads, social interaction correlates with lower efficiency in undisturbed cases and reduced responsiveness during encounters. An analysis of the impact parameter further shows that collision risk influences path deviations: individuals demonstrate larger deviations in response to highly interactive dyads, both in high-risk and less critical encounters. For dyads, the difference in deviation between low and high interaction levels is most pronounced when the single pedestrian is on a near-collision course. These results deepen our understanding of human pedestrian navigation, illustrating dynamical and social implications of group dynamics
Dependence of Perception of Vocabulary Difficulty on Contexture
No full textIn the collation and scheduling of learning material, it is common to collect feedback from learners about how confident they feel in remembering the items that they just studied. Based on such subjective opinion, the content of future learning tasks is decided such that the items which learners feel confident in remembering are reviewed after a relatively long interval and the ones which they do not feel confident are reviewed sooner. However, it is not clear how reliable is such an opinion or whether it can be biased by posing the inquiry in different modes. In this study, we focus on a learning scenario where non-native English speakers read passages, take a comprehension test and then evaluate a set of related vocabulary regarding their difficulty. We pose and test three hypotheses: (i) Learners' evaluations will be in line with the number of occurrences of the vocabulary (the more frequent the easier), should all options have the same lexical class, (ii) A word with an odd lexical class is likely to stick out, should all options be comparable in the number of occurrences and (iii) The option with the odd lexical class is likely to be ignored, should the remaining options have a clear ranking of the number of occurrences. In order to test our hypotheses, we made experiments with 10 participants. By assuming the independence of all observations, independence of all participants, and independence of all questionnaire items, we depict that the observed behavior has an underlying pattern that supports our claims in a statistically significant way
- …
