A5049012104- Evacuation and Crowd Dynamics
- Software Engineering Research
- Gaze Tracking and Assistive Technology
- Traffic and Road Safety
- Software Reliability and Analysis Research
- Traffic control and management
- Anomaly Detection Techniques and Applications
- Software Testing and Debugging Techniques
- Data Mining Algorithms and Applications
- Visual Attention and Saliency Detection
- Video Surveillance and Tracking Methods
- Software System Performance and Reliability
- Urban Design and Spatial Analysis
- Human Mobility and Location-Based Analysis
- Advanced Steganography and Watermarking Techniques
- Intelligent Tutoring Systems and Adaptive Learning
- Advanced Data Compression Techniques
- Big Data and Business Intelligence
- Imbalanced Data Classification Techniques
- Hand Gesture Recognition Systems
- EEG and Brain-Computer Interfaces
- Online Learning and Analytics
- Business Process Modeling and Analysis
- Rough Sets and Fuzzy Logic
- Subtitles and Audiovisual Media
Okayama University
2017-2024
Kyoto University
2024
Advanced Telecommunications Research Institute International
2010-2024
Osaka International University
2024
Okayama University of Science
2017-2023
City University
2022
Engineering (Italy)
2007-2019
Faculty (United Kingdom)
2007-2019
Bilkent University
2005-2009
University of Amsterdam
2009
The development of technologies for reliable tracking pedestrian trajectories in public spaces has recently enabled collecting large data sets and real-time information about the usage urban space indoor facilities by human crowds. Such an information, nevertheless, may be properly used only with aid theoretical computational tools to assess state crowd. As shown this work, traditional assessment metrics such as density flow provide a partial since it is also important understand how...
Joint attention, which is the ability of coordination a common point reference with communicating party, emerges as key factor in various interaction scenarios. This paper presents an image-based method for establishing joint attention between experimenter and robot. The precise analysis experimenter's eye region requires stability high-resolution image acquisition, not always available. We investigate regression-based interpolation gaze direction from head pose experimenter, easier to...
Being determined by human social behaviour, pedestrian group dynamics depends on "intrinsic properties" of the such as purpose pedestrians, their personal relation, gender, age, and body size. In this work we quantitatively study dynamical properties dyads analysing a large data set automatically tracked trajectories in an unconstrained "ecological" setting (a shopping mall), whose relational have been coded three different observers. We observed that females walk slower closer than males,...
This article presents a glossary of terms that are frequently used in research on human crowds. topic is inherently multidisciplinary as it includes work and across computer science, engineering, mathematics, physics, psychology social for example. We do not view the presented here collection finalised formal definitions. Instead, we suggest snapshot current views starting point an ongoing process hope will be useful providing some guidance use terminology to develop mutual understanding...
Head pose and eye location estimation are two closely related issues which refer to similar application areas. In recent years, these problems have been studied individually in numerous works the literature. Previous research shows that cylindrical head models isophote based schemes provide satisfactory precision estimation, respectively. However, locator is not adequate accurately locate presence of extreme poses. Therefore, cues may be suited enhance accuracy localization severe this...
Associating attributes to pedestrians in a crowd is relevant for various areas like surveillance, customer profiling and service providing. The of interest greatly depend on the application domain might involve such social relations as friends or family well hierarchy group including leader subordinates. Nevertheless, complex setting inherently complicates this task. We attack problem by exploiting small structures crowd. among individuals their peers within are reliable indicators...
This study focuses on social pedestrian groups in public spaces and makes an effort to identify the type of relation between group members. As a first step for this identification problem, we focus dyads (i.e. 2 people groups). Moreover, as mutually exclusive categorization relations, consider domain-based approach Bugental, which precisely corresponds relations colleagues, couples, friends families, each dyad with one those relations. For purpose, use anonymized trajectory data derive set...
In this work, we try to reproduce empirical results concerning the behaviour of a human crowd in cross-flow using hierarchy models, which differ details body shape (using disk-shaped vs more realistic elliptical shape) and how collision avoiding is performed only information regarding "centre mass" distance velocity, or actually introducing information). We verified that most detailed model (i.e., an body) outperforms significant way simplest one centre mass bodies). Furthermore, observed if...
In a follow-up to our work on the dependence of walking dyad dynamics intrinsic properties group, we now analyse how these affect groups three people (triads), taking also in consideration effect social interaction dynamical group. We show that there is strong parallel between triads and dyads. Work-oriented are faster walk at larger distance them than leisure-oriented ones, while latter move less ordered way. Such differences present when colleagues contrasted with friends families;...
In this work we investigate the behaviour of a human crowd in cross-flow by analysing results set controlled experiments which subjects were divided into two groups, organised such way to explore different density settings, and asked walk through crossing area. We study experiment defining investigating few macroscopic microscopic observables. Along with traditional indicators as velocity, whose dynamics was, extent our knowledge, poorly understood for setting, pay particular attention...
This paper elaborates on mechanisms for establishing visual joint attention the design of robotic agents that learn through natural interfaces, following a developmental trajectory not unlike infants. We describe first evolution cognitive skills in infants and then adaptation development patterns design. A comprehensive outlook cognitively inspired schemes pertaining to is presented last decade, with particular emphasis practical implementation issues. novel fixation mechanism defined agents.
Mobile social robots aimed at interacting with and assisting humans in pedestrian areas need to understand the dynamics of interaction. In this work, we investigate effect interaction on group motion by defining three models represent (1) interpersonal-distance, (2) relative orientation (3) absolute difference velocities; model them using a dataset 12000+ trajectories recorded uncontrolled settings. Our contributions include: (i) Demonstrating that has prominent empirical distributions...
The effectiveness of defect prediction depends on modeling techniques as well their parameter optimization, data preprocessing and ensemble development. This paper focuses auto-sklearn, which is a recently-developed software library for automated machine learning, that can automatically select appropriate models, hyperparameters given set develop with optimized weights. In this we empirically evaluate the auto-sklearn in predicting number defects modules. experiment, used metrics 20 OSS...
Pedestrian groups are commonly found in crowds but research on their social aspects is comparatively lacking. To fill that void literature, we study the dynamics of collision avoidance between pedestrian (in particular dyads) and individual pedestrians an ecological environment, focusing (i) how such depends group's relation (e.g. colleagues, couples, friends or families) (ii) its intensity interaction (indicated by conversation, gaze exchange, gestures etc). By analyzing relative "center...
This research centers on analyzing frontal encounters between dyads (two-person groups) and individuals, aiming to measure each participant's role in avoiding collisions based their deviation from intended path. To achieve this, we establish the trajectory of party by taking into account walking direction leading up encounter. The largest discrepancy this path observed can be interpreted as pedestrian's maximum lateral deviation.We show a noteworthy group members individuals face-to-face...
This study focuses on computer based foreign language vocabulary learning systems. Our objective is to automatically build decks with desired levels of relative difficulty relations. To realize this goal, we exploit the fact that word frequency a good indicator difficulty. Subsequently, for composing decks, pose two requirements as uniformity and diversity. Namely, level cards in same deck needs be uniform enough so they can grouped together different need diverse decks. assess diversity,...
This paper proposes a simple yet novel method for recognition of certain sorts moving entities incorporating their shape and motion patterns. Although features have been commonly employed in object recognition, characteristics are general not integrated to geometric models. In the interest utilizing attributes, trajectories investigated extract `coherence quality' entities. Besides, at every step model is adopted parameters defining utilized obtaining prior probabilities being member...
E-learning offers many advantages like being economical, flexible and customizable, but also has challenging aspects such as lack of – social-interaction, which results in contemplation sense remoteness. To overcome these sustain learners' motivation, various stimuli can be incorporated. Nevertheless, adjustments initially require an assessment engagement level. In this respect, we propose estimating level from facial landmarks exploiting the facts that (i) perceptual decoupling is promoted...
Modeling the user's attention is useful for responsive and interactive systems. This paper proposes a method establishing joint visual between an experimenter intelligent agent. A rapid procedure described to track 3D head pose of experimenter, which used approximate gaze direction. The modeled with sparse grid points sampled from surface cylinder. We then propose employ bottom-up saliency model single out interesting objects in neighborhood estimated focus attention. report results on...
Head pose and eye location estimation are two closely related issues which refer to similar application areas. In recent years, these problems have been studied individually in numerous works the literature. Previous research shows that cylindrical head models isophote based schemes provide satisfactory precision estimation, respectively. However, locator is not adequate accurately locate presence of extreme poses. Therefore, cues may be suited enhance accuracy localization severe this...