Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (other academic)  

Language：English   Publishing type：Research paper (scientific journal)  

Abstract: A method is proposed for resolving human congestion at a specific time at key spots in an area. Sensing data on real-world human flows are analyzed, and important information for changing movement behavior is accordingly provided. By using conventional approaches, this was a difficult task, whereas in the proposed approach, the targets and timing of providing information for congestion mitigation are determined based on spot importance. A congestion transition model is constructed from actual data and the results of a questionnaire survey. Finally, congestion mitigation in key spots is simulated after movement recommendation has been provided.

DOI： 10.3390/su12062417

Language：English   Publishing type：Research paper (other academic)  

This paper presents a method to predict three-dimensional (3D) plant growth with RGB-D images. Based on neural network based image translation and time-series prediction, we construct a system that gives the predicted result of RGB-D images from several past RGB-D images. Since both RGB and depth images are incorporated into our system, the plant growth can be represented in 3D space. In the evaluation, the performance of our proposed network is investigated by focusing on clarifying the importance of each module in the network. We have verified how the prediction accuracy changes depending on the internal structure of the our network.

Language：English   Publishing type：Research paper (other academic)  

The existing researches and developments of dashboard visualizing results from learning analytics mainly serve the instructors instead of learners in a direct manner. Effective visualizations extracted from learning log data can help the students to reflect and compare studying activities and access their metacognition to improve their self-regulated learning. For such purposes, we designed a reading path graph for visualizing the studying activities on slide pages used as teaching materials in classes intuitively, as one of the key functions of the learning dashboard. By providing the comparisons between the user’s own situation and the class overview, the visualization is expected to motivate the further actions of using other tools of the learning dashboard and reflecting studies. This paper introduces our exploration of the data process flows of extracting necessary data from a large number of operational logs for the visualization, and the techniques and strategies applied for rendering the graphics effectively. We implemented the data processing module with Python3 and the web-based visualization module of the reading path graph with JavaScript based on D3.js considering the extensibilities. The issues engaged in the development of prototypes are discussed, which will lead to the improvement of future prototypes and better designs of user experiments for formative evaluations as the next step of this research.

DOI： 10.1007/978-3-030-50344-4_41

Language：English   Publishing type：Research paper (other academic)  

Recently, various educational data such as clickstream data and eye movement data have been collected from students using e-learning systems. Learning analytics-based approaches also have been proposed such as student performance prediction and a monitoring system of student learning behaviors for supporting teachers. In this paper, we introduce our recent work as instances of the use of clickstream data and eye movement data. In our work, the clickstream data is used for representing student learning behaviors, and the eye movement data is used for estimating page areas where the student found difficulty. Besides, we discuss advantages and disadvantages depending on the types of educational data. To discuss them, we investigate a combination of highlights added on pages by students and eye movement data in page difficulty estimation. In the investigation, we evaluate the similarity between positions of highlights and page areas where the student found difficulty generated from eye movements. It is shown that areas in the difficult pages correspond to the highlights in this evaluation. Finally, we discuss how to combine the highlights and eye movement data.

DOI： 10.1007/978-3-030-50344-4_42

Language：English   Publishing type：Research paper (other academic)  

Semi-automated sensing and visualization of conditions and activities in farm fields have been actively pursued in recent years. There are three types of agricultural information: sensor information, farm work information, and plant biological information. Measuring and visualizing these agricultural information can provide valuable support to farm managers. In this study, we focus on sensor information and farm work information and develop a web application named SALATA (Sharing and AccumuLating Agricultural TAcit knowledge) that collects and shares sensor information and farm work information collected in farm fields and correlates the information in time series. SALATA need to have intuitive operation and quick response in order that people of various ages will use it on a daily basis. Therefore, there are two primary pages: the main page for visualizing simple information quickly and the analytical page for visualizing multiple pieces of information on one page. Usability evaluation experiments are performed, showing that SALATA can be operated intuitively and respond quickly.

Language：English   Publishing type：Research paper (scientific journal)  

This study presents an ontology-based visualization support system for e-book learners which promotes both meaningful receptive learning and meaningful discovery learning. To examine the system effectiveness, two learning modes are used: (a) reception comparison mode, where at the outset learners are shown complete versions of expert-generated topic maps; and (b) “cache-cache comparison mode,” where at the first stage of learning all information concerning relations is concealed, and at the second stage learners are encouraged to actively create those relations before comparing the learner-generated and expert-generated relations. The 50 control group participants studied in reception comparison mode while the 146 experimental groupparticipants studied in cache-cache comparison mode. Differences in learning perception and achievement between the two groups are examined, as is the effect of learner expertise level on learning mode effectiveness. Although the control group reported significantly more pressure and less satisfaction than the experimental group, no significant learning achievement differences were found between the two groups. However, in cache-cache comparison mode, the performance of learners with low prior knowledge increased more than that of learners with high prior knowledge; on the other hand, for learners with high prior knowledge, no significant effect of learning mode on learning achievement was found.

DOI： 10.1080/10494820.2020.1813178

Language：English   Publishing type：Research paper (other academic)  

Public datasets used to train modern object detection models do not contain all the object classes appearing in real-world surveillance scenes. Even if they appear, they might be vastly different. Therefore, object detectors implemented in the real world must accommodate unknown objects and adapt to the scene. We implemented a framework that combines background subtraction and unknown object detection to improve the pretrained detector’s performance and apply human intervention to review the detected objects to minimize the latent risk of introducing wrongly labeled samples to the training. The proposed system enhanced the original YOLOv3 object detector performance in almost all the metrics analyzed, and managed to incorporate new classes without losing previous training information.

Language：English   Publishing type：Research paper (other academic)  

This paper presents a ray-casting-based three-dimensional (3D) positioning system that interactively reconstructs scene structures for handheld augmented reality. The proposed system employs visual simultaneous localization and mapping (vSLAM) technology to acquire camera poses of a smartphone and sparse 3D feature points in an unknown scene. First, users specify a geometric shape region, such as a plane, in captured images while capturing a scene. This is performed by manually selecting some of the feature points generated by vSLAM in the region. Next, the system computes the shape parameter with the selected feature points so that the scene structure is reconstructed densely. Subsequently, users select the pixel of a target point in the scene at one camera view for 3D positioning. Finally, the system computes the intersection between the 3D ray computed with the selected pixel and the reconstructed scene structure to determine the 3D coordinates of the target point. Owing to the proposed interactive reconstruction, the scene structure can be estimated accurately and stably; therefore, 3D positioning will be accurate. Because the geometric shape used for the scene structure is a plane in this study, our system is referred to as PlanAR. In the evaluation, the performance of our system is compared statistically with an existing 3D positioning system to demonstrate the accuracy and stability of our system.

Language：English   Publishing type：Research paper (other academic)  

Localization is the process to compute sensor poses based on vision technologies such as visual Simultaneous Localization And Mapping (vSLAM). It can generally be applied to navigation systems . To achieve this, a global map is essential such that the relocalization process requires a single consistent map represented with an unified coordinate system. However, a large-scale global map cannot be created at once due to insufficient visual features at some moments. This paper presents an interactive method to generate a consistent global map from intermittent maps created by vSLAM independently via global reference points. First, vSLAM is applied to individual image sequences to create maps independently. At the same time, multiple reference points with known latitude and longitude are interactively recorded in each map. Then, the coordinate system of each individual map is converted into the one that has metric scale and unified axes with the reference points. Finally, the individual maps are merged into a single map based on the relative position of each origin. In the evaluation, we show the result of map merging and relocalization with our dataset to confirm the effectiveness of our method for navigation tasks. In addition, the report on participating in the navigation competition in a practical environment is also discussed.

Language：English   Publishing type：Research paper (scientific journal)  

The purpose of this study was to explore the factors that might affect learning performance and collaborative problem solving (CPS) awareness in science, technology, engineering, and mathematics (STEM) education. We collected and analyzed data on important factors in STEM education, including learning strategy and learning behaviors, and examined their interrelationships with learning performance and CPS awareness, respectively. Multiple data sources, including learning tests, questionnaire feedback, and learning logs, were collected and examined following a learning analytics approach. Significant positive correlations were found for the learning behavior of using markers with learning performance and CPS awareness in group discussion, while significant negative correlations were found for some factors of STEM learning strategy and learning behaviors in pre-learning with some factors of CPS awareness. The results imply the importance of an efficient approach to using learning strategies and functional tools in STEM education.

DOI： 10.1186/s41039-019-0119-y

Language：English   Publishing type：Research paper (other academic)  

In this paper, we present an ontology-based visualization support system for e-book learners, which provides not only a meaningful receptive learning environment but also a meaningful discovery learning environment. Those two environments are developed to help e-book learners to effectively construct their knowledge frameworks. A series of experiments were conducted on four undergraduate classes instructed by two professors (A and B): two classes(one guided by A and the other guided by B) were assigned as control groups and studied with one e-book chapter in receptive learning environment while another two classes (one guided by A and the other guided by B) were assigned as experimental groups and studied with the same e-book chapter in discovery learning environment. For analyzing the learner behavior, K-means clustering algorithm is performed not only by considering the number of total command actions and the cumulative duration of stay on target pages as learner features, but also by considering the duration of stay on each target page (in total 15 pages) as learner features. Learners’ behavior differences in e-book system are examined and discussed.

Language：English   Publishing type：Research paper (other academic)  

Learning log is defined as a digital record of what learners have learned in their daily lives using ubiquitous technologies. By using the ubiquitous learning system named SCROLL(System for Capturing and Remining Of Learning Logs), learners can save what they have learned in their daily lives with photo, such as location (latitude and longitude), learning place, and date and time of creation as a learning log. Although learners have many opportunities to learn words and meanings of objects with taking a photo in their daily lives, SCROLL is not implemented functions for supporting language learning with object and text detection. Therefore, this paper proposes a ubiquitous learning system to support language learning with object and text detection technologies.

Language：English   Publishing type：Research paper (other academic)  

Recently, background subtraction based on deep convolutional neural networks has demonstrated excellent performance in change detection tasks. However, most of the reported approaches require pixel-level label images for training the networks. To reduce the cost of rendering pixel-level annotation data, weakly supervised learning approaches using frame-level labels have been proposed. These labels indicate if a target class is present. Frame-level supervised learning is challenging because we cannot use location information for training the networks. Therefore, some constraints are introduced for guiding foreground locations. Previous works exploit prior information on foreground sizes and shapes. In this work, we propose two constraints for weakly supervised background subtraction networks. Our constraints use binary mask images generated by simple background subtraction. Unlike previous works, our approach does not require prior information on foreground sizes and shapes. Moreover, our constraints are more suitable for change detection tasks. We also present an experiment verifying that our constraints can improve foreground detection accuracy compared to other methods, which do not include them.

DOI： 10.1109/AVSS.2019.8909896

Language：English   Publishing type：Research paper (other academic)  

This paper describes a system that groups texts and objects in layout of slides. Layout of a slide is important to hold a presentation correctly. However, designing layout appropriately is difficult for beginners at making slides, like students. In addition, it is also difficult for them to check whether the layout of a slide is appropriate or not. To solve this problem, in this study, we developed a system that analyzes layout of a slide based on the positions of texts and shape objects. The system shows groups of texts each of which has common a topic. Beginners can realize mistakes of layout by comparing the output of the system and their intention.

DOI： 10.1109/IIAI-AAI.2019.00218

Language：English   Publishing type：Research paper (other academic)  

To improve designs of e-learning materials, it is necessary to know which word or figure a learner felt "difficult" in the materials. In this pilot study, we measured electroencephalography (EEG) and eye gaze data of learners and analyzed to estimate which area they had difficulty to learn. The developed system realized simultaneous measurements of physiological data and subjective evaluations during learning. Using this system, we observed specific EEG activity in difficult pages. Integrating of eye gaze and EEG measurements raised a possibility to determine where a learner felt “difficult” in a page of learning materials. From these results, we could suggest that the multimodal measurements of EEG and eye gaze would lead to effective improvement of learning materials. For future study, more data collection using various materials and learners with different backgrounds is necessary. This study could lead to establishing a method to improve e-learning materials based on learners' mental states.

DOI： 10.1145/3330430.3333648

Language：English   Publishing type：Research paper (other academic)  

Extended learning support systems for all-age education requires inclusive user interface design, especially for elderly users. A dual-tablet user interface with simplified visual layers and more intuitive operations was proposed aiming to reduce the physical and mental loads of elderly learners. An initial prototype with basic functions of viewing learning material was developed based on a cross-platform framework. Two preliminary user experiments participated by elderly volunteers were carried out for formative evaluations, in order to improve the usability of the interface design iteratively. The prototype was modified based on the participants’ comments and observation of their operations during the experiments. Additional findings of the elderly users’ preference and tendency were discussed for further development.

DOI： 10.1145/3330430.3333650

Language：English   Publishing type：Research paper (other academic)  

Extended learning support systems for all-age education requires inclusive user interface design, especially for elderly users. A dual-tablet user interface with simplified visual layers and more intuitive operations was proposed aiming to reduce the physical and mental loads of elderly learners. An initial prototype with basic functions of viewing learning material was developed based on a cross-platform framework. Two preliminary user experiments participated by elderly volunteers were carried out for formative evaluations, in order to improve the usability of the interface design iteratively. The prototype was modified based on the participants’ comments and observation of their operations during the experiments. Additional findings of the elderly users’ preference and tendency were discussed for further development.

DOI： 10.1145/3330430.3333650

Language：English   Publishing type：Research paper (other academic)  

To improve designs of e-learning materials, it is necessary to know which word or figure a learner felt "difficult" in the materials. In this pilot study, we measured electroencephalography (EEG) and eye gaze data of learners and analyzed to estimate which area they had difficulty to learn. The developed system realized simultaneous measurements of physiological data and subjective evaluations during learning. Using this system, we observed specific EEG activity in difficult pages. Integrating of eye gaze and EEG measurements raised a possibility to determine where a learner felt “difficult” in a page of learning materials. From these results, we could suggest that the multimodal measurements of EEG and eye gaze would lead to effective improvement of learning materials. For future study, more data collection using various materials and learners with different backgrounds is necessary. This study could lead to establishing a method to improve e-learning materials based on learners' mental states.

DOI： 10.1145/3330430.3333648

Language：English   Publishing type：Research paper (scientific journal)  

Online learning environments presently accumulate large amounts of log data. Analysis of learning behaviors from these log data is expected to benefit instructors and learners. This study was intended to identify effective measures from e-book materials used at Kyushu University and to employ these measures for analyzing learning behavioral patterns. In an evaluation, students were grouped into four clusters using k-means clustering, and their learning behavioral patterns were analyzed. We examined whether the learning behavioral patterns exhibited relations with the learning outcomes. The results reveal that the learning behavior of “backtrack” style reading exerts a significant positive influence on learning effectiveness, which can aid students to learn more efficiently.

DOI： 10.1080/10447318.2018.1543077

Language：English   Publishing type：Research paper (scientific journal)  

Fall is one of the leading causes of injury for the elderly individuals. Systems that automatically detect falls can significantly reduce the delay of assistance. Most of commercialized fall detection systems are based on wearable devices, which elderly individuals tend to forget wearing. Using surveillance cameras to detect falls based on computer vision is ideal, because anyone in the monitoring scopes can be under protection. However, the privacy protection issue using surveillance cameras has been bothering people. To effectively protect the privacy, we proposed an optical level anonymous image sensing system, which can protect the privacy by hiding the facial regions optically at the video capturing phase. We apply the system to fall detection. In detecting falls, we propose a neural network by combining a 3D convolutional neural network for feature extraction and an autoencoder for modelling the normal behaviors. The learned autoencoder reconstructs the features extracted from videos with normal behaviors with smaller average errors than those extracted from videos with falls. We evaluated our neural network by a hold-out validation experiment, and showed its effectiveness. In field tests, we showed and discussed the applicability of the optical level anonymous image sensing system for privacy protection and fall detection.

DOI： 10.1016/j.optlastec.2018.07.013

Language：English   Publishing type：Research paper (other academic)  

This paper introduces advanced tools in the digital learning management system M2B. The M2B system is used in Kyushu University, Japan, and contains three sub-systems: the e-learning system Moodle, the e-portfolio system Mahara, and the e-book system BookRoll. We developed useful tools to help improve both teaching and learning.

DOI： 10.1007/978-3-030-21935-2_32

Language：English   Publishing type：Research paper (other academic)  

Thanks to an increase in the amount of information on the Internet and the spread of ICT-supported educational environments, much attention has been paid to learning support based on "smart" recommendation technologies. In this study, we propose an education improvement model based on the recommender system using the human-in-the-loop design strategy. Our proposed model enhances not only learners via recommendation, but also teachers and the system itself through the interaction between teachers and the system. In this paper, we introduce the details of the proposed model and implementation strategy followed by a report of preliminary experimental results.

DOI： 10.33965/celda2019_201911l044

Language：English   Publishing type：Research paper (scientific journal)  

The Indoor Positioning and Indoor Navigation (IPIN) conference holds an annual competition in which indoor localization systems from different research groups worldwide are evaluated empirically. The objective of this competition is to establish a systematic evaluation methodology with rigorous metrics both for real-Time (on-site) and post-processing (off-site) situations, in a realistic environment unfamiliar to the prototype developers. For the IPIN 2018 conference, this competition was held on September 22nd, 2018, in Atlantis, a large shopping mall in Nantes (France). Four competition tracks (two on-site and two off-site) were designed. They consisted of several 1 km routes traversing several floors of the mall. Along these paths, 180 points were topographically surveyed with a 10 cm accuracy, to serve as ground truth landmarks, combining theodolite measurements, differential global navigation satellite system (GNSS) and 3D scanner systems. 34 teams effectively competed. The accuracy score corresponds to the third quartile (75^th percentile) of an error metric that combines the horizontal positioning error and the floor detection. The best results for the on-site tracks showed an accuracy score of 11.70 m (Track 1) and 5.50 m (Track 2), while the best results for the off-site tracks showed an accuracy score of 0.90 m (Track 3) and 1.30 m (Track 4). These results showed that it is possible to obtain high accuracy indoor positioning solutions in large, realistic environments using wearable light-weight sensors without deploying any beacon. This paper describes the organization work of the tracks, analyzes the methodology used to quantify the results, reviews the lessons learned from the competition and discusses its future.

DOI： 10.1109/ACCESS.2019.2944389

Language：English   Publishing type：Research paper (other academic)  

In recent times, researchers in pedagogy have focused on digital learning logs collected by learning tools such as Learning Management Systems and digital textbook systems. By analyzing and visualizing these, they aim to improve learning and/or teaching methodologies in the future. Using a digital textbook system, it is possible to collect information on which textbook pages were browsed by learners. However, these tools cannot decipher which zones of the textbook were browsed. In order to collect this information, eye-tracker technology would be necessary, but providing each learner with an eye-tracker would be too expensive. To solve this problem, a previous work proposed a method that used masks to detect and conceal each section of the slides in a digital textbook. The learner then clicked the masks one by one to delete them while browsing the contents of the digital textbook. By recording the learner’s clicking operations, the method collected information about the zones browsed by the learner. However, this method was found to cause a decline in learning achievement and system usability as a large number of zones were hidden. Therefore, we propose a grouping method, based on the layout information of the slides, in order to identify the appropriate zones to hide with masks.

DOI： 10.1007/978-3-030-21935-2_33

Language：English   Publishing type：Research paper (other academic)  

This paper presents a method for predicting plant growth in future images from past images, as a new phenotyping technology. This is achieved by modeling the representation of plant growth based on neural network. In order to learn the long-term dependencies in plant growth from the images, we propose to employ a Convolutional LSTM based framework. Especially, We apply an encoder-decoder model inspired by a framework on future frame prediction to model the representation of plant growth effectively. In addition, we propose two additional loss terms to put the constraints on shape changes of leaves between consecutive images. In the evaluation, we demonstrated the effectiveness of the proposed loss functions through the comparisons using labeled plant growth images.

Language：English   Publishing type：Research paper (other academic)  

In this paper, we focus on optimizing the assignment of students to courses. The target courses are conducted by different teachers using the same syllabus, course design, and lecture materials. More than 1,300 students are mechanically assigned to one of ten courses taught by different teachers. Therefore, mismatches often occur between students' learning behavior patterns and teachers' approach to teaching. As a result, students may be less satisfied, have a lower level of understanding of the material, and achieve less. To solve these problems, we propose a strategy to optimize the assignment of students to courses based on learning activity analytics. The contributions of this study are 1) clarifying the relationship between learning behavior pattern and teaching based on learning activity analytics using large-scale educational data, 2) optimizing the assignment of students to courses based on learning behavior pattern analytics, and 3) demonstrating the effectiveness of assignment optimization via simulation experiments.

Language：English   Publishing type：Research paper (scientific journal)  

We propose a multi-pedestrian tracking system based on MEMS based IMUs as a novel tool for human behavior analysis. With asynchronized multiple IMUs, our system can track IMU-attached pedestrians in synchronization at a high frame rate in the large environment, compared with vision based approaches. The output data is similar to standard PDR systems as follows: the time-series position, velocity, and heading of the pedestrians in the 3D space. To realize our system, we propose a simple but effective calibration technique for synchronizing the timelines of the asynchronized IMUs. With our system, users can analyze the detailed motion behaviors of the people who participate in a group work or a collective activity, quantitatively. By combining with other sensors such as an eye tracker, our system can further provide more comprehensive data in the experiments.

Language：English   Publishing type：Research paper (other academic)  

Thanks to an increase in the amount of information on the Internet and the spread of ICT-supported educational environments, much attention has been paid to learning support based on "smart" recommendation technologies. In this study, we propose an education improvement model based on the recommender system using the human-in-the-loop design strategy. Our proposed model enhances not only learners via recommendation, but also teachers and the system itself through the interaction between teachers and the system. In this paper, we introduce the details of the proposed model and implementation strategy followed by a report of preliminary experimental results.

Language：English   Publishing type：Research paper (other academic)  

This study investigates how we can understand students' actual status in C programming exercises from their learning activity logs. In a face-to-face course of C programming exercise, it is hard for a teacher to see who are in trouble from their apperance. It is not always true that typing something means he or she is making some progress. Therefore it is important to identify, or possibly even predict, students having difficulty from their activity patterns. Most of the prior work paid attention to only trial-and-error activities, such as compile results and execution errors. However, it tends to be overlooked that knowledge acquisition process is also worthy of attention. When a student encounters a compile error, they usually read textbooks to seek a solution. It is considered to be useful for the task whether he or she has an ability to find appropriate pages for error resolution. In this paper, we propose a method to predict whether a student can resolve errors or not. Based on students' activity logs collected from our programming environment and e-book system, we conduct experiments to show and discuss the prediction performance.

Language：English   Publishing type：Research paper (other academic)  

Extended learning environments involving system to collect data for learning analytics and to support learners will be useful for all-age education. As the first steps towards to build new learning environments, we developed a system for multimodal learning analytics using eye-tracker and EEG measurement, and inclusive user interface design for elderly learners by dual-tablet system. Multimodal learning analytics system can be supportive to extract where and how learners with varied backgrounds feel difficulty in learning process. The eye-tracker can retrieve information where the learners paid attention. EEG signals will provide clues to estimate their mental states during gazes in learning. We developed simultaneous measurement system of these multimodal responses and are trying to integrate the information to explore learning problems. A dual-tablet user interface with simplified visual layers and more intuitive operations was designed aiming to reduce the physical and mental loads of elderly learners. A prototype was developed based on a cross-platform framework, which is being refined by iterative formative evaluations participated by elderlies, in order to improve the usability of the interface design. We propose a system architecture applying the multimodal learning analytics and the user-friendly design for elderly learners, which couples learning analytics “in the wild” environment and learning analytics in controlled lab environments.

DOI： 10.1007/978-3-030-22580-3_35

Language：English   Publishing type：Research paper (other academic)  

This paper presents an outline of our project, in which we develop an observation framework for integrating lecture and contextual learning in the field of crop cultivation. Specifically, we will use multi sensing of learners' activities in classrooms, and contextual learning in fieldwork, farm planting, and farming environments. The motivation for our project is twofold: First, crop cultivation provides a powerful illustration of educational technology. It requires both explicit knowledge (from lectures) and implicit knowledge (from contextual learning outside of class). Second, from a practical viewpoint, the number of Japanese farmers is shrinking due to low income and to aging population. Thus, in order to maintain crop yields, farming skills must be transferred efficiently to novice farm workers. Herein, the major features of our framework will be described.

Language：English   Publishing type：Research paper (scientific journal)  

Demand for indoor navigation systems has been rapidly increasing with regard to location-based services. As a cost-effective choice, inertial measurement unit (IMU)-based pedestrian dead reckoning (PDR) systems have been developed for years because they do not require external devices to be installed in the environment. In this paper, we propose a PDR system based on a chest-mounted IMU as a novel installation position for body-suit-type systems. Since the IMU is mounted on a part of the upper body, the framework of the zero-velocity update cannot be applied because there are no periodical moments of zero velocity. Therefore, we propose a novel regression model for estimating step lengths only with accelerations to correctly compute step displacement by using the IMU data acquired at the chest. In addition, we integrated the idea of an efficient map-matching algorithm based on particle filtering into our system to improve positioning and heading accuracy. Since our system was designed for 3D navigation, which can estimate position in a multifloor building, we used a barometer to update pedestrian altitude, and the components of our map are designed to explicitly represent building-floor information. With our complete PDR system, we were awarded second place in 10 teams for the IPIN 2018 Competition Track 2, achieving a mean error of 5.2 m after the 800 m walking event.

DOI： 10.3390/s19020420

Language：English   Publishing type：Research paper (other academic)  

With the development of green energy and its means of production, more and more companies chose to build solar panel farms. However, those technologies remain relatively expensive to maintain, and prone to damages (due to natural hazards, or internal defects). Since any kind of damage on a panel cell drastically reduce a panel's efficiency, solar panels must be kept under tight supervision. With more solar panel that must be checked for damage relatively often, a cheap, accurate and fast way to find those damages must be settled. Some processes have been developed to identify panels in a true color image [1], and various ways to identify defective panels exist through image processing [2], [3] or other ways [4]. On another hand, handmade features suggest the input data obeys to some specific conditions (color, illumination), and small changes can impact accuracy. CNN [5], however, can be trained to face such changes with the appropriate dataset, and therefore be more resilient. They represent a reliable solution for identification and classification of complex features [2], [6], and can be improved more easily than handmade feature detection. In this paper is detailed the pipeline of such process, combining the straightforward approach of handmade feature detection for preprocessing to reduce the input's complexity, with the resilience of neural networks for the final identification. Detailed explanations for the different steps of the process are given: Dataset acquisition, preprocessing, and finally classification. The various leads that were followed to improve the quality of the results are also given, before comparing results with a previously used handmade detection process, and finally proposing a web user interface to exploit this process, and enrich its dataset.

DOI： 10.1117/12.2522098

Language：English   Publishing type：Research paper (scientific journal)  

This paper describes a method to collect data of which section of pages learners were browsing in digital textbooks without eye-tracking technologies. In previous researches on digital textbook systems, it was difficult to collect such data without using eye-tackers. However, eye-trackers cost a massive budget. Our proposed system automatically hides the texts in the digital textbooks with mask processing before the learners browse the texts in the digital textbooks. If they click the hidden texts, the system gets rid of the masks and the texts appear letter by letter. We used NMF to discover learners’ browsing patterns from the collected logs. Evaluation experiments were conducted to examine the effectiveness of our system in terms of fascination, understandableness and enhancement of thinking and to discover learners’ browsing patterns. It was found that our method could enhance thinking skills. A browsing pattern of diligent learners with high learning achievements was also found.

DOI： 10.1080/10494820.2019.1619594

Language：English   Publishing type：Research paper (other academic)  

This study aimed to cluster learners based on the structures of the knowledge maps they created. Learners drew their own knowledge maps to reflect their learning activities. Our system collected individual knowledge maps from many learners and clustered them to generate an integrated version of the knowledge maps of each cluster. We applied the graph analysis method to extract important keywords from the knowledge map. The results of the analysis showed that the utilization of the knowledge map helped to improve lectures and grasp the learners' level of understanding. We conducted surveys asking course managers to evaluate the effectiveness of the integrated knowledge maps of learners included in the cluster and received both positive and negative responses.

Language：English   Publishing type：Research paper (scientific journal)  

This paper describes an automatic quiz generation system designed to support language learning that utilizes digital textbook logs. Learners often memorize words in digital textbooks while preparing for an examination, and they often use the highlight function for the words. Previous studies regarding annotations and highlights have shown that learning only by using the highlight function on important content in textbooks did not affect learning achievements. Therefore, in this study, we developed a system that can support the repeated learning by analyzing digital textbook logs and providing appropriate quizzes. An evaluation experiment involving 31 international students was conducted to assess whether the quizzes provided by our proposed system are able to enhance the learning achievements as compared to teacher-created quizzes. The results show that the quizzes by our proposed system and the teacher-created quizzes were both equally effective. A correlation analysis was conducted to identify the correlation among the learning achievements, the number of quizzes, and each variable in questionnaires. We found that there is a positive correlation between the number of quizzes and the students’ learning achievements.

DOI： 10.1080/10494820.2019.1620291

Language：English   Publishing type：Research paper (scientific journal)  

Purpose: The purpose of this study is to propose a real-time lecture supporting system. The target of this study is on-site classrooms where teachers give lectures and a lot of students listen to teachers’ explanations, conduct exercises, etc. Design/methodology/approach: The proposed system uses an e-learning system and an e-book system to collect teaching and learning activities from a teacher and students in real time. The collected data are immediately analyzed to provide feedback to the teacher just before the lecture starts and during the lecture. For example, the teacher can check which pages were well previewed and which pages were not previewed by students using the preview achievement graph. During the lecture, real-time analytics graphs are shown on the teacher’s PC. The teacher can easily grasp students’ status and whether or not students are following the teacher’s explanation. Findings: Through the case study, the authors first confirmed the effectiveness of each tool developed in this study. Then, the authors conducted a large-scale experiment using a real-time analytics graph and investigated whether the proposed system could improve the teaching and learning in on-site classrooms. The results indicated that teachers could adjust the speed of their lecture based on the real-time feedback system, which also resulted in encouraging students to put bookmarks and highlights on keywords and sentences. Originality/value: Real-time learning analytics enables teachers and students to enhance their teaching and learning during lectures. Teachers should start considering this new strategy to improve their lectures immediately.

DOI： 10.1108/ITSE-05-2018-0026

Language：English   Publishing type：Research paper (scientific journal)  

This paper proposes a new approach to visualizing spatial variation of plant status in a tomato greenhouse based on farm work information operated by laborers. Farm work information consists of a farm laborer’s position and action. A farm laborer’s position is estimated based on radio wave strength measured by using a smartphone carried by the farm laborer and Bluetooth beacons placed in the greenhouse. A farm laborer’s action is recognized based on motion data measured by using smartwatches worn on both wrists of the farm laborer. As experiment, harvesting information operated by one farm laborer in a part of a tomato greenhouse is obtained, and the spatial distribution of yields in the experimental field, called a harvesting map, is visualized. The mean absolute error of the number of harvested tomatoes in each small section of the experimental field is 0.35. An interview with the farm manager shows that the harvesting map is useful for intuitively grasping the states of the greenhouse.

DOI： 10.3390/s18113906

Language：English   Publishing type：Research paper (scientific journal)  

Stereo matching has been solved as a supervised learning task with convolutional neural network (CNN). However, CNN based approaches basically require huge memory use. In addition, it is still challenging to find correct correspondences between images at ill-posed dim and sensor noise regions. To solve these problems, we propose Sparse Cost Volume Net (SCV-Net) achieving high accuracy, low memory cost and fast computation. The idea of the cost volume for stereo matching was initially proposed in GC-Net. In our work, by making the cost volume compact and proposing an efficient similarity evaluation for the volume, we achieved faster stereo matching while improving the accuracy. Moreover, we propose to use weight normalization instead of commonly-used batch normalization for stereo matching tasks. This improves the robustness to not only sensor noises in images but also batch size in the training process. We evaluated our proposed network on the Scene Flow and KITTI 2015 datasets, its performance overall surpasses the GC-Net. Comparing with the GC-Net, our SCV-Net achieved to: (1) reduce 73.08% GPU memory cost; (2) reduce 61.11% processing time; (3) improve the 3PE from 2.87% to 2.61% on the KITTI 2015 dataset.

DOI： 10.3390/rs10111844

Language：English   Publishing type：Research paper (other academic)  

This paper describes an infrastructure for seamless learning analytics to bridge digital textbook learning and practical training such as programming and conceptual modeling education. To realize the infrastructure for seamless learning analytics, we propose the integration of a digital textbook system into a software learning support system. By using our proposed infrastructure, all learning data will be sent by xAPI and collected in an independent LRS. We believe that analyzing and visualizing the relationships between the learning in the digital textbook system and practical training in the software learning support system leads to improving the quality of learning and teaching.

Language：English   Publishing type：Research paper (other academic)  

This paper proposes a new method of early change detection for people flow analysis. Some conventional methods often focus on a single location (spot) to demonstrate how the number of people changes over time. In contrast, our proposed method takes into account the links between the spots to grasp a foretaste of congestion of a specific spot as early as possible. The main advantage of the proposed method is that it not only describes the characteristics of each spot, but also the relationships among spots, i.e., whether the connectivities are strong/weak. We introduce an idea of PageRank, which is based on a centrality of graph theory and extend that idea to represent the amount of people flow among spots. We call the extended method “SpotRank”. SpotRank assigns an importance score to each spot. The score of a particular spot is calculated by the number of paths and the amount of people flow from other spots. Therefore, the more paths and people flow, the importance score (ranking) increases. The proposed method begins with the calculation of SpotRank every 10 min, followed by change detection, i.e., how much the ranking changes over time. In our experiments, we measured people flow using Wi-Fi packet sensors for a period of over 16 weeks. We confirmed the effectiveness of the proposed method, which successfully grasped a foretaste of congestion at a restaurant in our university.

DOI： 10.1145/3267305.3267565

Language：English   Publishing type：Research paper (other academic)  

This paper describes redesign of data collection in digital textbook systems. Previous studies in the digital textbook systems allow us to collect the data which pages learners were browsing in the digital textbooks. However, it is difficult to collect the data which positions of the pages learners were browsing in the digital textbooks unless using eye-tracking technologies. This study developed a digital textbook system called SEA (Smart E-textbook Application) to collect the data which positions of the pages the learners were browsing in the digital textbooks without eye-tracking technologies. The system automatically hides the texts in the digital textbook system with mask processing. An evaluation experiment was conducted to evaluate the user acceptance regarding our developed system.

DOI： 10.1109/IIAI-AAI.2018.00192

Language：English   Publishing type：Research paper (scientific journal)  

This paper is an extension of research originally reported in [1]. Here, we propose a novel method for summarizing lecture slides to enhance students' preview efficiency and understanding of the content. Students are often asked to prepare for a class by reading lecture materials. However, because the attention span of students is limited, this is not always beneficial. We surveyed 326 students regarding the preview of lecture materials, revealing a preference for summarized materials to preview. Therefore, we developed an automatic summarization method for condensing original lecture materials into a summarized set. Our proposed approach utilizes image and text processing to extract important pages from lecture materials, optimizing selection of pages in accordance with a specified preview time. We applied the proposed summarization method to a set of lecture slides. In an experiment with 372 students, we compared the effectiveness of the summarized slides and the original materials in terms of quiz scores, preview achievement ratio, and time spent previewing. We found that students who previewed the summarized slides achieved better scores on pre-lecture quizzes, even though they spent less time previewing the material.

DOI： 10.1109/TLT.2017.2682086

Language：English   Publishing type：Research paper (scientific journal)  

Reconstruction-based change detection methods are robust for camera motion. The methods learn reconstruction of input images based on background images. Foreground regions are detected based on the magnitude of the difference between an input image and a reconstructed input image. For learning, only background images are used. Therefore, foreground regions have larger differences than background regions. Traditional reconstruction-based methods have two problems. One is over-reconstruction of foreground regions. The other is that decision of change detection depends on magnitudes of differences only. It is difficult to distinguish magnitudes of differences in foreground regions when the foreground regions are completely reconstructed in patch images. We propose the framework of a reconstruction-based change detection method for a free-moving camera using patch images. To avoid over-reconstruction of foreground regions, our method reconstructs a masked central region in a patch image from a region surrounding the central region. Differences in foreground regions are enhanced because foreground regions in patch images are removed by the masking procedure. Change detection is learned from a patch image and a reconstructed image automatically. The decision procedure directly uses patch images rather than the differences between patch images. Our method achieves better accuracy compared to traditional reconstruction-based methods without masking patch images.

DOI： 10.3390/s18041232

Language：English   Publishing type：Research paper (other academic)  

We propose a new change detection method using clickstream data collected through an e-Book system. Most of the prior work has focused on the batch processing of clickstream data. In contrast, the proposed method is designed for online processing, with the model parameters for change detection updated sequentially based on observations of new click events. More specifically, our method generates a model for an individual student and performs minute-by-minute change detection based on click events during a classroom lecture. We collected clickstream data from four face-to-face lectures, and conducted experiments to demonstrate how the proposed method discovered change points and how such change points correlated with the students’ performances.

DOI： 10.1145/3170358.3170412

Repository Public URL： https://hdl.handle.net/2324/7341548

Language：English   Publishing type：Research paper (other academic)  

We introduce our sensing technologies to acquire agricultural information, such as image-based plant phenotyping, harvest quantity data, and localization information using a camera in a greenhouse. Commercial systems exist that support agriculture, but many unresolved issues remain regarding optimization of farming sustainability and productivity. Therefore, we intend to apply state-of-the-art information and communication technology (ICT) to tackle these agricultural issues and to investigate their limitations for developing advanced smart agricultural systems.

DOI： 10.1109/ICSensT.2017.8304451

Language：English   Publishing type：Research paper (other academic)  

This preliminary study developed the concept map tool “BR-Map” using learning logs on eBook viewer, and investigated the relationships between self-regulated learning (SRL) awareness, learning behaviors (usage of BR-Map, and one-minute paper and report submission), and learning performance. Psychometric data and learning logs were collected in the lecture course, and their relationships were analyzed using Spearman’s correlation analysis. The results indicated that awareness of intrinsic value, use of cognitive learning strategies, and self-regulation had significant correlations with the usage of BR-Map. The awareness of cognitive learning strategies had significant correlation with standard deviation of one-minute paper submission hours. With regard to relationships between the BR-Map usage and learning behaviors, the relationships between the usage of BR-Map and one-minute paper submissions, which was a regularly weekly assigned task, were found.

Language：English   Publishing type：Research paper (other academic)  

Our research goal is to construct a system to measure farm labor activities in a farm field and visualize farm field information based on the activities. As the first step for the goal, this paper proposes a method to measure harvesting information of farm labors in a tomato greenhouse and to visualize the tomato yield distribution in the greenhouse, we call it a harvesting map, for supporting the farm managers making decisions. A harvesting map shows daily, weekly and monthly tomato yields in small sections into which the tomato greenhouse is divided.

DOI： 10.1007/978-3-319-91125-0_16

Language：English   Publishing type：Research paper (other academic)  

We discuss the applicability of a fully convolutional network (FCN), which provides promising performance in semantic segmentation tasks, to plant segmentation tasks. The challenge lies in training the network with a small dataset because there are not many samples in plant image datasets, as compared to object image datasets such as ImageNet and PASCAL VOC datasets. The proposed method is inspired by transfer learning, but involves a two-step adaptation. In the first step, we apply transfer learning from a source domain that contains many objects with a large amount of labeled data to a major category in the plant domain. Then, in the second step, category adaptation is performed from the major category to a minor category with a few samples within the plant domain. With leaf segmentation challenge (LSC) dataset, the experimental results confirm the effectiveness of the proposed method such that F-measure criterion was, for instance, 0.953 for the A2 dataset, which was 0.355 higher than that of direct adaptation, and 0.527 higher than that of non-adaptation.

DOI： 10.5220/0006576303320339

Language：English   Publishing type：Research paper (other academic)  

This paper describes a learning analytics perspective for designing to implement a seamless learning environment. Seamless learning has been focused on supporting learning across formal and informal learning contexts, individual and social learning and physical world and cyberspace. Majority of the current researches have realized a seamless learning environment by using the technologies such as smart-phone and GPS at schools or universities. However, utilization of the collected learning logs still remains a challenge yet to be explored. In this study, to construct a seamless learning environment, this study developed a system that integrated a digital textbook system called AETEL with a ubiquitous learning system called SCROLL. The system enables learners to bridge the learning between digital textbook learning and real-life learning. To analyze and visualize the relationships between them, this study developed an innovative system called VASCORLL 2.0 (Visualization and analysis System for Connecting Relationships of Learning Logs). An experiment was conducted to evaluate whether VASCORLL 2.0 can increase learners’ learning opportunities. As a result, they were able to increase their learning opportunities by using VASCORLL 2.0. It contributed to enhancing learning activities in the seamless learning environment by utilizing the collected learning logs with well-designed analysis and visualization approaches.

DOI： 10.1007/978-3-319-91131-1_8

Language：English   Publishing type：Research paper (scientific journal)  

Deep neural network-based (DNN-based) background subtraction has demonstrated excellent performance for moving object detection. The DNN-based background subtraction automatically learns the background features from training images and outperforms conventional background modeling based on handcraft features. However, previous works fail to detail why DNNs work well for change detection. This discussion helps to understand the potential of DNNs in background subtraction and to improve DNNs. In this paper, we observe feature maps in all layers of a DNN used in our investigation directly. The DNN provides feature maps with the same resolution as that of the input image. These feature maps help to analyze DNN behaviors because feature maps and the input image can be simultaneously compared. Furthermore, we analyzed important filters for the detection accuracy by removing specific filters from the trained DNN. From the experiments, we found that the DNN consists of subtraction operations in convolutional layers and thresholding operations in bias layers and scene-specific filters are generated to suppress false positives from dynamic backgrounds. In addition, we discuss the characteristics and issues of the DNN based on our observation.

DOI： 10.3390/jimaging4060078

Language：Japanese   Publishing type：Research paper (scientific journal)  

Lecture Support based on Real-time Learning Analytics

Language：English   Publishing type：Research paper (other academic)  

Group work is widely introduced and practiced as a method to achieve the learning goal efficiently by collaborating group members. However, since most types of group works are carried out in the real environment, it is very difficult to perform formative assessment and real time evaluation without students’ feedbacks. Therefore, there is a strong demand to develop a method that supports evaluation of group work. To support evaluation of group work, this paper proposes a method to visualize the real world activity during group work by using first person view cameras and wearable sensors. Here, the proposed method visualizes three scores: (1) individual attention, (2) hand visibility, (3) individual activity. To evaluate the performance and analyze the relationships between scores, we conducted experiments of “Marshmallow challenge” that is a collaborative work to construct a tower using marshmallow and spaghetti within a limit of time. Through the experiments, we confirmed that the proposed method has potential to become a evaluation tool for visualizing the activity of the group work.

DOI： 10.1007/978-3-319-91131-1_2

Language：English   Publishing type：Research paper (other academic)  

As smart technologies pervade our everyday environments, they change what people should learn to live meaningfully as valuable participants of our society. For instance, ubiquitous availability of smart devices and communication networks may have reduced the burden for people to remember factual information. At the same time, they may have increased the benefits to master the uses of new digital technologies. In the midst of such a social and technological shift, we could design novel integrated platforms that support people at all ages to learn, work, collaborate, and co-create easily. In this paper, we discuss our ideas and first steps towards building an extended learning analytics platform that elderly people and unskilled adults can use. By understanding the characteristics and needs of elderly learners and addressing critical user interface issues, we can build pervasive and inclusive learning analytics platforms that trigger contextual reminders to support people at all ages to live and learn actively regardless of age-related differences of cognitive capabilities. We discuss that resolving critical usability problems for elderly people could open up a plethora of opportunities for them to search and exploit vast amount of information to achieve various goals.

DOI： 10.1007/978-3-319-91131-1_6

Language：English   Publishing type：Research paper (other academic)  

We propose a method of energy management aimed at reducing the emission of carbon dioxide by changing people’s behavior in small and medium-sized electricity communities. In the conventional energy management system, a power peak is cut and shifted mainly using solar power generation and batteries. In this research, a power peak is cut and shifted by controlling the power demand. The power demand for each facility in small communities is controlled by changing crowd behavior. In experiments, models for predicting power demand according to crowd congestion are constructed for each facility and the accuracies of prediction are verified.

DOI： 10.1007/978-3-319-91125-0_20

Language：English   Publishing type：Research paper (other academic)  

The paper discusses the potential of wearable devices and mainly focuses on smartwatches for super-aging societies. A smartwatch is designed to be worn on the wrist similar to a traditional watch. Additionally, smartwatches involve touch screens, software applications, and IMU (inertial measurement unit) sensors. Smartphones are mainly utilized for notifications related to phone calls, mails, SNS, or healthcare applications to measure and record heart rate and other vital signals. Conversely, smartwatches exhibit significant potential to generate new value in our daily life. In the paper, we introduce three applications with smartwatches published in extant studies and discuss the applicability of smartwatch applications to super-aging societies.

DOI： 10.1007/978-3-319-91131-1_17

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we explore the factor for improving the performance of prediction of students’ quiz scores by using a Recurrent Neural Network. The proposed method is applied to the log data of 2693 students in 15 courses that were conducted with following the common syllabus by 10 teachers. The experimental results show that in the case where the same teacher is not included in both training and test data, the accuracy of prediction slightly lower. We also show that at the beginning of a course, it is better to construct a prediction model including various items of learning logs, however, in the latter half, it is better to update the model by using selected information only.

Language：English   Publishing type：Research paper (other academic)  

Many studies related to congestion analysis focus on estimating quantitative values such as actual number of people, mobile devices, and crowd density. In contrast, we focus on perceptual congestion rather than quantitative congestion; however, we also analyze the relationship between quantitative and perceptual congestion. We construct a system for estimating and visualizing congestion and collecting user reports about congestion. We use the number of mobile devices as quantitative congestion measurements obtained from Wi-Fi packet sensors and a user report-based congestion as a perceptual congestion measurement collected via our Web system. In our experiments, we investigate the relationship between these values. In addition, we apply Non-negative Tensor Factorization to extract latent patterns between locations and congestion. These latent features help us to understand the relationship of the characteristics among the locations.

DOI： 10.1007/978-3-319-93647-5_12

Language：English   Publishing type：Research paper (scientific journal)  

A thermal camera captures the temperature distribution of a scene as a thermal image. In thermal images, facial appearances of different people under different lighting conditions are similar. This is because facial temperature distribution is generally constant and not affected by lighting condition. This similarity in face appearances is advantageous for face detection. To detect faces in thermal images, cascade classifiers with Haar-like features are generally used. However, there are few studies exploring the local features for face detection in thermal images. In this paper, we introduce two approaches relying on local features for face detection in thermal images. First, we create new feature types by extending Multi-Block LBP. We consider a margin around the reference and the generally constant distribution of facial temperature. In this way, we make the features more robust to image noise and more effective for face detection in thermal images. Second, we propose an AdaBoost-based training method to get cascade classifiers with multiple types of local features. These feature types have different advantages. In this way we enhance the description power of local features. We did a hold-out validation experiment and a field experiment. In the hold-out validation experiment, we captured a dataset from 20 participants, comprising 14 males and 6 females. For each participant, we captured 420 images with 10 variations in camera distance, 21 poses, and 2 appearances (participant with/without glasses). We compared the performance of cascade classifiers trained by different sets of the features. The experiment results showed that the proposed approaches effectively improve the performance of face detection in thermal images. In the field experiment, we compared the face detection performance in realistic scenes using thermal and RGB images, and gave discussion based on the results.

DOI： 10.3390/s17122741

Language：English   Publishing type：Research paper (scientific journal)  

This research aims to investigate the relationship between self-regulated learning awareness, learning behaviors, and learning performance in ubiquitous learning environments. In order to conduct this research, psychometric data about self-regulated learning and log data, such as slide pages that learners read, marker, and annotate, was collected. The accessing activity of device types that stored the learning management system was collected and analyzed by applying path analysis and correlation analysis using data divided into high and low performers. The results indicated that the slide pages which learners read for a duration of between 240 and 299 s had positive effects on the promotion of annotation and the learning performance directly, and albeit indirectly, the enhancement of self-efficacy was affected by other self-regulated learning factors. The results of the correlation analysis indicated that self-efficacy and test anxiety are a key factor that has different effects on the number of the read slide pages in both high and low performers.

DOI： 10.1186/s41039-017-0053-9

Language：English   Publishing type：Research paper (other academic)  

Recently, deep neural networks (DNNs) have demonstrated excellent performance for change detection. The DNN-based background subtraction automatically discovers background features from datasets and outperforms traditional background modeling based on handcraft features and/or subtraction strategies. Most researchers mainly discuss the accuracy of foreground detection and do not analyze how and why the DNN works well for change detection tasks. It is necessary to understand what the DNN learns as background features in order to discuss the potential of the DNN in background subtraction. In this paper, we focus on the filters in the first convolution layer and the activations of neurons in the last fully connected layer to understand the behavior of the DNN. From the experiment, we found that 1) the first layer performs the role of background subtraction using several filters, and 2) the last layer categorizes some background changes into a group without supervised signals. These findings suggest the possibility of a new background modeling strategy based on data-driven extracted features.

DOI： 10.1109/AVSS.2017.8078550

Language：English   Publishing type：Research paper (scientific journal)  

We propose a framework for adaptively registering background models with an image for background subtraction with moving cameras. Existing methods search for a background model using a fixed window size, to suppress the number of false positives when detecting the foreground. However, these approaches result in many false negatives because they may use inappropriate window sizes. The appropriate size depends on various factors of the target scenes. To suppress false detections, we propose adaptively controlling the method parameters, which are typically determined heuristically. More specifically, the search window size for background registration and the foreground detection threshold are automatically determined using the re-projection error computed by the homography based camera motion estimate. Our method is based on the fact that the error at a pixel is low if it belongs to background and high if it does not. We quantitatively confirmed that the proposed framework improved the background subtraction accuracy when applied to images from moving cameras in various public datasets.

DOI： 10.1016/j.patrec.2017.03.010

Language：English   Publishing type：Research paper (other academic)  

To discover teaching knowledge efficiently, we must extract the various teaching activities from educational data. In this paper, through the use of e-book logs and techniques of time-series analysis, we describe a method of practicing teaching analytics in face-to-face classes, one which enable us to extract the teaching activity efficiently and accurately.

DOI： 10.1109/ICALT.2017.75

Language：English   Publishing type：Research paper (other academic)  

Students' reflective writings are useful not only for students themselves but also teachers. It is important for teachers to know which concepts were understood well by students and which concepts were not, to continuously improve their classes. However, it is difficult for teachers to thoroughly read the journals of more than one hundred students. In this paper, we propose a novel method to extract common topics and students' common impressions against them from students' journals. Weekly keywords are discovered from journals by scoring noun words with a measure based on TF-IDF term weighting scheme, and then we analyze co-occurrence relationships between extracted keywords and adjectives. We employs nonnegative matrix factorization, one of the topic modeling techniques, to discover the hidden impression topics from the co-occurrence relationships. As a case study, we applied our method on students' journals of the course 'Information Science' held in our university. Our experimental results show that conceptual keywords are successfully extracted, and four significant impression topics are identified. We conclude that our analysis method can be used to collectively understand the impressions of students from journal texts.

DOI： 10.1109/ICALT.2017.113

Language：English   Publishing type：Research paper (other academic)  

A real-time learning analytics system is proposed for in-classroom use. We used an e-learning system and an e-book system to collect real-time learning activities during lectures. The collected logs were analyzed and presented visually on a web-based system for the teacher. The teacher can monitor how many students are viewing the same page as the teacher, whether they are following the explanation, or if they are reading previous or subsequent pages. Through a case study, we confirmed the effectiveness of the real-time learning analytics system, in terms of high synchronization between the teacher and the students, i.e., that the majority of students followed the teacher's explanation and added more bookmarks, highlights, or notes on the e-book, compared with the control group where the teacher did not use our system.

DOI： 10.1109/ICALT.2017.74

Language：English   Publishing type：Research paper (other academic)  

We present a 3D phenotyping platform that measures both plant growth and environmental information in small indoor environments for plant image datasets. Our objective is to construct a compact and complete platform by using commercial devices to allow any researcher to begin plant phenotyping in their laboratory. In addition, we introduce our annotation tool to manually but effectively create leaf labels in plant images on a pixel-by-pixel basis. Finally, we show our RGB-D and multiview datasets containing images in the early growth stages of the Komatsuna with leaf annotation.

DOI： 10.1109/ICCVW.2017.239

Language：English  

DOI： 10.1007/978-3-319-55345-0_13

Language：English   Publishing type：Research paper (other academic)  

In this paper, we propose a method for predicting final grades of students by a Recurrent Neural Network (RNN) from the log data stored in the educational systems. We applied this method to the log data from 108 students and examined the accuracy of prediction. From the experimental results, comparing with multiple regression analysis, it is confirmed that an RNN is effective to early prediction of final grades.

DOI： 10.1145/3027385.3029479

Language：English   Publishing type：Research paper (other academic)  

In this paper, we examined whether previous findings on educational big data consisting of e-book logs from a given academic course can be reproduced with different data from other academic courses. The previous findings showed that (1) students who attained consistently good achievement more frequently browsed different e-books and their pages than low achievers and that (2) this difference was found only for logs of preparation for course sessions (preview), not for reviewing material (review). Preliminarily, we analyzed e-book logs from four courses. The results were reproduced in only one course and only partially, that is, (1) high achievers more frequently changed e-books than low achievers (2) for preview. This finding suggests that to allow effective usage of learning and teaching analyses, we need to carefully construct an educational environment to ensure reproducibility.

DOI： 10.1145/3027385.3029445

Language：English   Publishing type：Research paper (other academic)  

Many universities choose the C programming language (C) as the first one they teach their students, early on in their program. However, students often consider programming courses difficult, and these courses often have among the highest dropout rates of computer science courses offered. It is therefore critical to provide more effective instruction to help students understand the syntax of C and prevent them losing interest in programming. In addition, homework and paper-based exams are still the main assessment methods in the majority of classrooms. It is difficult for teachers to grasp students' learning situation due to the large amount of evaluation work. To facilitate teaching and learning of C, in this article we propose a system-LAPLE (Learning Analytics in Programming Language Education)-that provides a learning dashboard to capture the behavior of students in the classroom and identify the different difficulties faced by different students looking at different knowledge. With LAPLE, teachers may better grasp students' learning situation in real time and better improve educational materials using analysis results. For their part, novice undergraduate programmers may use LAPLE to locate syntax errors in C and get recommendations from educational materials on how to fix them.

DOI： 10.1145/3027385.3027407

Language：English   Publishing type：Research paper (other academic)  

A new lecture supporting system based on real-time learning analytics is proposed. Our target is on-site classrooms where teachers give their lectures, and a lot of students listen to teachers' explanation, conduct exercises etc. We utilize not only an e-Learning system, but also an e-Book system to collect real-time learning activities during the lectures. The proposed system is useful for a teacher just before lecture starts and during the lecture. The system provides summary reports of the previews of given materials and quiz results. The teacher can check which pages were well previewed and which pages were not previewed by students using the preview achievement graph. Additionally, the teacher can check which quizzes were difficult for students, and the suggested pages that should be explained in the lecture to aid students. During the lecture, real-time analytics graphs are shown on the teacher's PC. The teacher can easily grasp students status whether or not students are following the teacher's explanation. Through a case study, we confirmed the effectiveness of the proposed system, in terms of high synchronization between a teacher and students, i.e., the teacher adjusted the speed of his lecture based on the real-time feedback, and many students followed the teacher's explanation.

Language：English   Publishing type：Research paper (other academic)  

This paper describes a real-time learning analytics to find learning contents or keywords that students don't understand in digital textbooks. We developed a digital textbook viewer system that can collect students' learning logs. By analyzing and visualizing the collected learning logs in real time, teachers can visually find the keywords that students don't understand during a class. This paper describes the contribution of real-time learning analytics for supporting teachers.

Language：English   Publishing type：Research paper (other academic)  

This paper discusses gesture design for man-machine interfaces. Traditionally, gesture-interface studies have focused on improving performance, in terms of increasing speed and accuracy, in particular reducing false positives. Many studies neglect to consider the gestures’ intrinsic machine friendliness, which can improve recognition accuracy, and user friendliness, which makes a gesture easier to use and to remember. In this paper, we investigate machine-and user-friendly gestures and analyze the results of an Internet-based questionnaire in which 351 individuals were asked to assign gestures to eight operations.

DOI： 10.1007/978-3-319-51811-4_35

Language：English   Publishing type：Research paper (other academic)  

In this paper, we show a method to predict students' final grades using a recurrent neural network (RNN). An RNN is a variant of a neural network that handles time series data. For this purpose, the learning logs from 937 students who attended one of six courses by two teachers were collected. Nine kinds of learning logs are selected as the input of the RNN. We examine the prediction of final grades, where the training data and test data are the logs of courses conducted in 2015 and in 2016, respectively. We also show a way to identify the important learning activities for obtaining a specific final grade by observing the values of weight of the trained RNN.

Language：English   Publishing type：Research paper (other academic)  

In foreground segmentation, the depth information is robust to problems of the appearance information such as illumination changes and color camouflage; however, the depth information is not always measured and suffers from depth camouflage. In order to compensate for the disadvantages of the two pieces of information, we define an energy function based on the two likelihoods of depth and appearance backgrounds and minimize the energy using graph cuts to obtain a foreground mask. The two likelihoods are obtained using background subtraction. We use the farthest depth as the depth background in the background subtraction according to the depth information. The appearance background is defined as the appearance with a large likelihood of the depth background to eliminate appearances of foreground objects. In the computation of the likelihood of the appearance background, we also use the likelihood of the depth background for reducing false positives owing to illumination changes. In our experiment, we confirm that our method is sufficiently accurate for indoor environments using the SBM-RGBD 2017 dataset.

DOI： 10.1007/978-3-319-70742-6_25

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we address the problem of social image tagging using practical vocabulary for mobile users on the social media. On the social media, images usually have an incomplete or noisy set of social tags provided by the mobile users, and we consider this issue as defective tag assignments. Previous studies on social image tagging have mostly focused on multi-label classification without considering the defective tags. In these studies, the usage of multi-label classification techniques is expected to synergically exploit the linear relations between the image features and the semantic tags. However, these approaches usually aimed to capture the linear relations from the training data while ignoring the helpful information from the test data. In addition, they failed to incorporate the non-linear associations residing in the visual features as well as in the semantic tags. To overcome these drawbacks, we introduce a novel approach based on non-linear matrix completion for image tagging task with defective tags. Specifically, we first construct the entire feature-tag matrix based on the visual features with non-linear kernel mapping. Then, we present a formal methodology together with an optimization method under the matrix completion framework to jointly complete the tags of training and test images. Experimental evaluations demonstrate that our method shows promising results on image tagging task on two benchmark social image datasets with defective tags, and establishes a baseline for such models in this research domain.

DOI： 10.1109/ACCESS.2016.2624267

Language：English   Publishing type：Research paper (other academic)  

An infrared (IR) camera captures the temperature distribution of an object as an IR image. Because facial temperature is almost constant, an IR camera has the potential to be used in detecting facial regions in IR images. However, in detecting faces, a simple temperature thresholding does not always work reliably. The standard face detection algorithm used is AdaBoost with local features, such as Haar-like, MB-LBP, and HOG features in the visible images. However, there are few studies using these local features in IR image analysis. In this paper, we propose an AdaBoost-based training method to mix these local features for face detection in thermal images. In an experiment, we captured a dataset from 20 participants, comprising 14 males and 6 females, with 10 variations in camera distance, 21 poses, and participants with and without glasses. Using leave-one-out cross-validation, we show that the proposed mixed features have an advantage over all the regular local features.

DOI： 10.1117/12.2266836

Language：English   Publishing type：Research paper (scientific journal)  

We investigated the learning behavior of undergraduates with e-book logs. E-book logs from 99 undergraduates taking an information science course were collected. First, we analyzed differences between nine high-achieving students and three low-achieving students. A log recorded before a class session in which the same e-book was used as a textbook was considered a preview log, and one recorded after a class session was considered a review log. The analysis of preview frequency indicates that the low achievers did not perform the previews, but many high achievers frequently did. The review frequency demonstrates that regardless of high and low achievements, students performed reviews. We added the logs of six relatively low achievers and analyzed more details of the preview logs of high and low achievers. The number of page flips and durations of preview logs revealed that relatively low achievers tried to perform previews, but they gave the endeavor up easily.

Language：English   Publishing type：Research paper (other academic)  

Although people flow analysis is widely studied because of its importance, there are some difficulties with previous methods, such as the cost of sensors, person re-identification, and the spread of smartphone applications for collecting data. Today, Probe Request sensing for people flow analysis is gathering attention because it conquers many of the difficulties of previous methods. We propose a framework for Probe Request data analysis for extracting the latent behavior patterns of people. To make the extracted patterns understandable, we apply a Non-negative Tensor Factorization with a sparsity constraint and initialization with prior knowledge to the analysis. Experimental result showed that our framework helps the interpretation of Probe Request data.

DOI： 10.5220/0006193901570164

Language：English   Publishing type：Research paper (other academic)  

Students' journal writings could be useful resources for teachers to grasp their understandings and to see their own teaching objectively. However, reading a large number of journals thoroughly is not always realistic for teachers. Although various automatic analysis methods have been proposed to understand learning journals, they does not necessarily fit needs of teachers and tend to overlook minor opinions. In this paper, we propose an interactive report tool for exploring journal writings. Focusing on the efficiency of reading learning journals, it employs weekly keywords extracted from journals as entry points for journal sentences. It enables us to read journal sentences selectively. The tool also provides lists of most used adjectives from week to week, which is helpful for teachers to grasp the temporal variation of opinions through a semester. We conducted a preliminary questionnaire about the usefulness of the report tool targeting teachers of the course "Information Science" in our university. Most of them evaluated our tool positively although the number of answers were small.

Language：English   Publishing type：Research paper (other academic)  

In this paper, we propose a cross analytics methodology of student activities and course activities using e-Book operation logs collected in 15 courses with face-to-face lecture style over 4 weeks. These courses commonly use the same lecture materials, but are conducted by different teachers. The new aspect of our research is that we perform cross analysis over courses. Most past researches focus on students' activities in a specific course, and give discussions about how the students behaved, how the behaviors differ from each other. In contrast, our research focuses on the course activities and conducts a comparison among courses. First, we begin with data alignment for row data to rectify a student activity every 10 seconds. Through our analytics, it becomes clear that whether students' activities varies with teachers or their teaching styles. In the experiments, we applied the proposed analytics to 1.1-million operation logs, and found out interesting characteristics through the comparison across courses.

Language：English   Publishing type：Research paper (other academic)  

E-book has been gradually getting popularity in educational contexts. Reading textbooks on computers or hand-held devices enables us to track the learning activities of students regardless of situations. In our university, several courses for first year students employs our e-book system, and we have been collecting its usage logs. From the logs, it seems that the highlighter function of the e-book reader plays an important role in learning because it is used most by the students. Though many researches studied the effectiveness of e-textbooks, only limited studies addressed how students utilize highlighters and how marking activity affects their learning. In this paper, we focus on highlighted portions of e-textbooks, and analyze how students use highlighters in their learning. We also attempt to provide recommendations to students for highlighting based on the highlighter usage in other classes.

Language：English   Publishing type：Research paper (other academic)  

Conventional works for congestion estimates focus on estimating quantitative congestion (e.g., actual number of people, mobile devices, and crowd density). Meanwhile, we focus on perceptual congestion rather than quantitative congestion toward providing perceptual congestion information. We analyze the relationship between quantitative and perceptual congestion. For this analysis, we construct a system for estimating and visualizing congestion and collecting user reports about congestion. We use the number of mobile devices as quantitative congestion measurements obtained from Wi-Fi packet sensors, and user-report-based congestion as a perceptual congestion measurement collected via our Web service. Base on the obtained quantitative and perceptual congestion, we investigate the relationship between these values.

DOI： 10.5220/0006206102250232

Language：English   Publishing type：Research paper (other academic)  

In this paper, we present a system for visualizing learning logs of a course in progress together with predictions of learning activities of the following week and the final grades of students by state transition graphs. Data are collected from 236 students attending the course in progress and from 209 students attending the past course for prediction. From these data, the system constructs a state transition graph, where the prediction is based on the Markov property. We verify the performance of predictions by experiments in which the accuracy of prediction using the data of the course in progress and the one by 5-fold cross validation.

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, a global algorithm for facial and gesture recognition is presented. The algorithm basically consists of three modules: features sensing and processing, dominant features selection and finally features matching. Depending on the type of data used (vision or sensor based), the proposed algorithm exploits multiple features employing 2DPCA that efficiently compact features’ descriptors maintain the spatial and temporal alignment of features’ components. Canonical Correlation Analysis (CCA) is employed to fuse different features from different descriptors or different performers. CCA also transforms training and testing features sets into new space where similar pairs become highly correlated pairs. Different experiments were conducted using well known data sets in addition to our newly collected data sets to verify the efficiency of the proposed algorithm. Excellent recognition accuracy, and fast performance are factors that promotes the proposed algorithm for real time implementation.

DOI： 10.1007/s40860-016-0028-4

Language：English   Publishing type：Research paper (other academic)  

This paper describes visualization and analysis methods using educational big data collected by research project at Kyushu University in Japan. The project uses an e-book system called BookLooper, Moodle, and Mahara. Logs for this analytics were collected from 99 first-year students in an information science course at Kyushu University. The number of logs are collected approximately 330,000, and this paper visualize and analyze the collected logs. The purpose of this study is to predict students' final grade and to profile visualization and analysis results. The prediction of this study shows that it leads to discoveries of students who fail to make the grade.

DOI： 10.1109/ICALT.2016.27

Language：English   Publishing type：Research paper (scientific journal)  

Nowadays the amount of multimedia data such as images and text is growing exponentially on social websites, arousing the demand of effective and efficient cross-modal retrieval. The cross-modal hashing based methods have attracted considerable attention recently as they can learn efficient binary codes for heterogeneous data, which enables large-scale similarity search. Generally, to effectively construct the cross-correlation between different modalities, these methods try to find a joint abstraction space where the heterogeneous data can be projected. Then a quantization rule is applied to convert the abstraction representation to binary codes. However, these methods may not effectively bridge the semantic gap through the latent abstraction space because they fail to capture latent information between heterogeneous data. In addition, most of these methods apply the simplest quantization scheme (i.e. sign function) which may cause information loss of the abstraction representation and result in inferior binary codes. To address these challenges, in this paper, we present a novel cross-modal hashing based method that generates unified binary codes combining different modalities. Specifically, we first extract semantic features from the modalities of images and text to capture latent information. Then these semantic features are projected to a joint abstraction space. Finally, the abstraction space is rotated to produce better unified binary codes with much less quantization loss, while preserving the locality structure of projected data. We integrate the binary code learning procedures above to develop an iterative algorithm for optimal solutions. Moreover, we further exploit the useful class label information to reduce the semantic gap between different modalities to benefit the binary code learning. Extensive experiments on four multimedia datasets show that the proposed binary coding schemes outperform several other state-of-the-art methods under cross-modal scenarios.

DOI： 10.1016/j.neucom.2015.11.133

Language：English   Publishing type：Research paper (scientific journal)  

A new early gesture recognition system that uses different features obtained from MYO sensor is presented. The beginning part of each gesture is detected and used by the system to train the authors' recognition algorithm. To preserve the different features temporal alignment for each movement, two-dimensional (2D) principal component analysis was employed to obtain the dominant features by processing the obtained data in its 2D form. Canonical correlation analysis (CCA) is used to find a space where the projection of similar training testing pairs becomes highly correlated. Finally, the testing sequence is matched to the training set that gives maximum correlation in the new space obtained by CCA. Low processing complexity, storage requirement, accurate and fast decision obtained on the newly collected data set are factors that promotes the authors' algorithm for real-time implementation.

DOI： 10.1049/el.2016.1540

Language：English   Publishing type：Research paper (scientific journal)  

This paper is an extension of the work that was originally reported in Shimada et al. (2013). This paper proposes a change detection method based on spatio-temporal light ray consistency. The proposed method introduces light field sensing, which is used to generate an arbitrary in-focus plane. Change detection is performed in a surveillance scene, where the background region can be filtered out by an out-focusing process. This approach resolves a longstanding issue in background modeling-based object detection, which often suffers from false positives in the background regions. To realize this new change detection method, a new feature representation, called the local ray pattern (LRP), is introduced. The LRP evaluates the spatial consistency of the light rays, and this plays an important role in distinguishing whether the light rays come from the in-focus plane or elsewhere. A combination of the LRP and Gaussian mixture model (GMM)-based background modeling realizes change detection in the in-focus plane. Experimental results demonstrate the proposed method's effectiveness and its applicability to video surveillance.

DOI： 10.1016/j.jvcir.2016.02.013

Language：English   Publishing type：Research paper (other academic)  

This paper presents a new gesture recognition algorithm that uses different features obtained from MYO sensor. To preserve the spatial and temporal alignment for different features of each movement, Two Dimensional Principal Component Analysis 2DPCA is employed to obtain the dominant features by processing the obtained data in its 2D form. Canonical Correlation Analysis CCA is used to find a space where the projection of similar training/testing pairs become highly correlated. The testing sequences is matched to the training set that gives maximum correlation in the new space obtained by CCA. Two new data sets for common HCI applications (gaming and air writing) were collected at LIMU lab, Kyushu university and used to verify the efficiency of the proposed algorithm. Low processing complexity, efficient storage requirement, high accuracy and fast decision are factors that promotes our algorithm for real time implementation.

DOI： 10.1109/MWSCAS.2016.7870154

Language：English   Publishing type：Research paper (scientific journal)  

The goal of image annotation is to automatically assign a set of textual labels to an image to describe the visual contents thereof. Recently, with the rapid increase in the number of web images, nearest neighbor (NN) based methods have become more attractive and have shown exciting results for image annotation. One of the key challenges of these methods is to define an appropriate similarity measure between images for neighbor selection. Several distance metric learning (DML) algorithms derived from traditional image classification problems have been applied to annotation tasks. However, a fundamental limitation of applying DML to image annotation is that it learns a single global distance metric over the entire image collection and measures the distance between image pairs in the image-level. For multi-label annotation problems, it may be more reasonable to measure similarity of image pairs in the label-level. In this paper, we develop a novel label prediction scheme utilizing multiple label-specific local metrics for label-level similarity measure, and propose two different local metric learning methods in a multi-task learning (MTL) framework. Extensive experimental results on two challenging annotation datasets demonstrate that 1) utilizing multiple local distance metrics to learn label-level distances is superior to using a single global metric in label prediction, and 2) the proposed methods using the MTL framework to learn multiple local metrics simultaneously can model the commonalities of labels, thereby facilitating label prediction results to achieve state-of-the-art annotation performance.

DOI： 10.1007/s11042-014-2402-7

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we propose a novel method to make a summary set of lecture slides for supporting students' review study. Quizzes are often conducted in a lecture to check students' understanding level. The aim of our study is to support a student who wrongly answers the quiz. The quiz statement is analyzed to extract nouns in the statement. Then, text mining is performed to find the pages related to the quiz statement in the relevant lecture materials. The proposed SummaryRank algorithm evaluates the topic similarity among pages in material with emphasizing the related page to the quiz statement. In addition, our proposed method considers the preview status of each student, resulting in the generation of adaptive review materials tailored for each student. Through experiments, we confirmed that the proposed method could find appropriate pages with respect to the quiz statements.

Language：English   Publishing type：Research paper (scientific journal)  

DOI： 10.3390/systems4020021

Language：English   Publishing type：Research paper (other academic)  

In this paper, a video object segmentation method utilizing the motion of superpixel centroids is proposed. Our method achieves the same advantages of methods based on clustering point trajectories, furthermore obtaining dense clustering labels from sparse ones becomes very easy. Simply for each superpixel the label of its centroid is propagated to all its entire pixels. In addition to the motion of superpixel centroids, histogram of oriented optical flow, HOOF, extracted from superpixels is used as a second feature. After segmenting each object, we distinguish between foreground objects and the background utilizing the obtained clustering results.

DOI： 10.1007/978-3-319-41501-7_22

Language：English   Publishing type：Research paper (scientific journal)  

The purpose of this paper is to mine or detect meaningful learning patterns for profiling high-achieving students using e-book-based activity logs and questionnaire. The analysis of this study uses association analysis with Apriori algorithm. Logs for this analysis were collected from 99 first-year students who use a document viewer system called BookLooper, questionnaires and Moodle in an information science course at Kyushu University. From the results of the association analysis, we found that high-achieving students and BookLooer have significant relationships in terms of preparation and review time. This paper believes that the profiling and analysis can be used to predict their final grades and to detect effective learning patterns.

Language：English   Publishing type：Research paper (other academic)  

This research aims to investigate the relationship between self-regulated learning awareness, learning behaviors, and learning performance in ubiquitous learning environments. In order to do so, psychometric data about self-regulated learning and log data such as marker, annotation, accessing device types that stored the learning management system were collected and analyzed using multiple regression analysis with stepwise method. The results indicated that self-efficacy, internal value, and the number of read slides had a significant influence on the final score, and the awareness of cognitive learning strategy use has slightly significant power to predict the final score.

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we present a method of identifying learning activities that are important for students to achieve good grades. For this purpose, the data of 99 students were collected from a learning management system and an e-book system, including attendance, time on preparation and review, submission of reports, and quiz scores. We applied a support vector machine to these data to calculate a score of importance for each learning activity reflecting its contribution to the attainment of an A grade. Selecting certain important learning activities by following several evaluation measures, we verified that these learning activities played a crucial role in predicting final student achievements. One of the obtained results implies that time on preparation and review in the middle part of a course influences a student's final achievement.

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we address the problem of image annotation with incomplete labeling, where multiple objects in each training image are not fully labeled. The conventional one-versus-all support vector machine (OVA-SVM), which performs fairly well on full labeling, decays drastically under the setting of incompleteness. Recently, a structured output learning method termed OVA-SSVM was proposed to boost the performance of OVA-SVM by modeling the structured associations of labels and show efficiency under the setting of incompleteness. OVA-SSVM assumes that each training sample includes a single label and adopts an loss measure of classification style where, as long as one of the predicted label is correct, the overall prediction should be considered correct. However, this may not be appropriate for the multilabel annotation task. Therefore, we extend the OVA-SSVM method to the multilabel situation and design a novel image-specific structured loss to account for the dependences between predicted labels relying on image label associations. The superiority of the proposed image-specific structured loss is that it can directly learn the semantic relationships of labels from training data without predefined semantic hierarchy. Extensive empirical results on a variety of benchmark datasets show that the proposed method performs significantly better than OVA-SSVM on image annotation tasks with incomplete labeling and achieves competitive performance compared to other state-of-the-art methods.

DOI： 10.1002/tee.22190

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we propose a background model by using an exponentially weighted updating method. We realize to reduce memory usage for construction of background model. Our background model is represented as a histogram according to pixel values. Our model uses an exponential increasing weight for updating our model. In our model, recently observed pixels have a bigger influence on the background model than older ones. Therefore, our model gradually ignores the effect of old-observed value on a background model without retaining past pixel values. We apply our method to background subtraction for comparing with conventional methods using kernel density estimation. In experiments, we conformed that the detection accuracy of our background model is comparable to that of conventional methods.

Language：English   Publishing type：Research paper (other academic)  

As smartwatches are becoming more widely used in society, gesture recognition, as an important aspect of interaction with smartwatches, is attracting attention. An accelerometer that is incorporated in a device is often used to recognize gestures. However, a gesture is often detected falsely when a similar pattern of action occurs in daily life. In this paper, we present a novel method of designing a new gesture that reduces false detection. We refer to such a gesture as a low-false-positive (LFP) gesture. The proposed method enables a gesture design system to suggest LFP motion gestures automatically. The user of the system can design LFP gestures more easily and quickly than what has been possible in previous work. Our method combines primitive gestures to create an LFP gesture. The combination of primitive gestures is recognized quickly and accurately by a random forest algorithm using our method. We experimentally demonstrate the good recognition performance of our method for a designed gesture with a high recognition rate and without false detection.

DOI： 10.5220/0005701905810588

Language：English   Publishing type：Research paper (other academic)  

Background modeling and subtraction are essential to video surveillance applications. There are two main issues related to background modeling: how to initialize the background model, and how to update the model based on observations. In this paper, we consider the first issue with the aim of generating a clear background image that does not contain foreground objects or noise. We used a bidirectional analysis and consensus voting strategy to achieve this goal. We demonstrated the effectiveness of our technique using open access datasets.

DOI： 10.1109/ICPR.2016.7899620

Language：English   Publishing type：Research paper (other academic)  

We propose a strategy of background subtraction for an image sequence captured by a moving camera. To adapt for camera motion, it is necessary to estimate the relation between consecutive frames in background subtraction. However, simple background subtraction using the relation between consecutive frames results in many false detections. We use re-projection error to handle this problem. The re-projection error has a low value in a background region. According to re-projection error, our method searches neighboring background models and tunes a threshold value for detection in order to reduce false detections. We evaluated the accuracy of detection of our method in experiments. Our method provided better detection than a method that does not search neighboring background models. Our method thus reduced the number of false detections.

DOI： 10.1109/ICIP.2015.7351278

Language：English   Publishing type：Research paper (other academic)  

Existing background model based change detection methods have difficulty in distinguishing between foreground and background changes when both changes are caused by the same factors. We explore the possibility of using a light field camera to resolve the problem of existing single-view camera-based approaches. We present a new change detection strategy that processes light rays captured by the light field camera. The light rays are used for three purposes: 1) generating an active surveillance field (ASF) to determine in-focus and out-focus areas, 2) evaluating focusness to determine whether the light rays come from the ASF, and 3) creating and updating light-ray background models to capture temporal changes in light rays. To investigate the effectiveness of the proposed approach, we evaluated several video sequences captured by a light field camera. Experimental results show that our change detection scheme can robustly handle challenging situations that cannot be resolved by existing single-view approaches.

DOI： 10.1109/AVSS.2015.7301785

Language：English   Publishing type：Research paper (other academic)  

Nowadays massive amount of images and texts has been emerging on the Internet, arousing the demand of effective cross-modal retrieval. To eliminate the heterogeneity be-tween the modalities of images and texts, the existing sub-space learning methods try to learn a common latent sub-space under which cross-modal matching can be performed. However, these methods usually require fully paired sam-ples (images with corresponding texts) and also ignore the class label information along with the paired samples. In-deed, the class label information can reduce the semantic gap between different modalities and explicitly guide the subspace learning procedure. In addition, the large quan-tities of unpaired samples (images or texts) may provide useful side information to enrich the representations from learned subspace. Thus, in this paper we propose a novel model for cross-modal retrieval problem. It consists of 1) a semi-supervised coupled dictionary learning step to generate homogeneously sparse representations for different modali-ties based on both paired and unpaired samples; 2) a coupled feature mapping step to project the sparse representations of different modalities into a common subspace defined by class label information to perform cross-modal matching. Exper-iments on a large scale web image dataset MIRFlickr-1M with both fully paired and unpaired settings show the effec-tiveness of the proposed model on the cross-modal retrieval task.

DOI： 10.1145/2733373.2806346

Language：English   Publishing type：Research paper (other academic)  

In this paper, we present a visualization tool for person re-identification when tracking objects across non-overlapping cameras. Tracking objects across non-overlapping cameras is challenging because the observations from different cameras are widely separated in both time and space. Hence, these systems need a large amount of labeled training data. Commonly, this training data is constructed manually at significant human cost. We support this process efficiently by visualizing the correspondences of objects across multiple cameras. Our tool facilitates the construction of a database for person re-identification with ease. Moreover, the accuracy of person re-identification can be increased using the generated database because the amount of training data is increased. In the experiments, we apply the proposed tool to real world situations to verify the validity of the proposed system.

DOI： 10.1109/AVSS.2015.7301740

Language：English   Publishing type：Research paper (other academic)  

This paper discusses learning behavior analysis using a learning management system (LMS) and an e-textbook system. We collected a large number of operation logs from e-textbooks to analyze the process of learning. In addition, we conducted a quiz to check the level of understanding. In our study, we especially focus on an analysis of the relationship between learning behavior in informal learning and its effectiveness in the corresponding quiz. We apply a machine learning and classification methodology for behavior analysis. Our experimental results demonstrate that students who undertake good informal learning achieve better scores in quizzes.

DOI： 10.1109/ICALT.2015.78

Language：English   Publishing type：Research paper (other academic)  

This preliminary research investigates the relationship between psychometric data and learning behaviors in the learning analytics research field, specifically, the relationship between self-regulated learning and learning behavior. The results of this limited research show that marker and annotation use have a weak significant relationship with self-efficacy and the intrinsic value of learning materials.

DOI： 10.1109/ICALT.2015.74

Language：English   Publishing type：Research paper (other academic)  

In this paper, we investigate the problem of modeling images and associated text for cross-modal retrieval tasks such as text-to-image search and image-to-text search. To make the data from image and text modalities comparable, previous cross-modal retrieval methods directly learn two projection matrices to map the raw features of the two modalities into a common subspace, in which cross-modal data matching can be performed. However, the different feature representations and correlation structures of different modalities inhibit these methods from efficiently modeling the relationships across modalities through a common subspace. To handle the diversities of different modalities, we first leverage the coupled dictionary learning method to generate homogeneous sparse representations for different modalities by associating and jointly updating their dictionaries. We then use a coupled feature mapping scheme to project the derived sparse representations from different modalities into a common subspace in which cross-modal retrieval can be performed. Experiments on a variety of cross-modal retrieval tasks demonstrate that the proposed method outperforms the state-of-the-art approaches.

DOI： 10.1109/ICME.2015.7177396

Language：English   Publishing type：Research paper (scientific journal)  

Local features, such as scale-invariant feature transform (SIFT) and speeded up robust features (SURF), are widely used for describing an object in the applications of visual object recognition and classification. However, these approaches cannot apply to transparent objects made of glass or plastic, as such objects take on the visual features of background scenes, and the appearance of such objects dramatically varies with changes in the scenes. Indeed, transparent objects have the unique characteristic of distorting the background by refraction. In this paper, we use a single-shot light field image as input and model the distortion of the light field caused by the refractive property of a transparent object. We propose a new feature which is called the light field distortion (LFD) feature. The proposed feature is background-invariant so that it is able to describe a transparent object without knowing the texture of the scene. The proposal incorporates this LFD feature into the bag-of-features approach for classifying transparent objects. We evaluated its performance and analyzed the limitations in various settings.

DOI： 10.1016/j.cviu.2015.02.009

Language：English   Publishing type：Research paper (other academic)  

We present spatially-multiplexed fiducial markers with the framework of code division multiple access (CDMA), which is a technique in the field of communications. Since CDMA based multiplexing is robust to signal noise and interference, multiplexed markers can be demultiplexed under several image noises and transformation. With this framework, we explore the paradigm of multiple-input and multiple-output (MIMO) for fiducial markers so that the data capacity of markers can be improved and different users can receive different data from a multiplexed marker.

DOI： 10.1109/3DUI.2015.7131765

Language：English   Publishing type：Research paper (other academic)  

Detection of moving objects is one of the key steps for vision based applications. Many previous works leverage background subtraction using background models and assume that image sequences are captured from a stationary camera. These methods are not directly applied to image sequences from a moving camera because both foreground and background objects move with respect to the camera. One of the approaches to tackle this problem is to estimate background movement by computing pixel correspondences between frames such as homography. With this approach, moving objects can be detected by using existing background subtraction. In this paper, we evaluate detection of foreground objects for image sequences from a moving camera. Especially, we focus on homography as a camera motion. In our evaluation we change the following parameters: changing feature points, the number of them and estimation methods of homography. We analyze its effect on detection of moving objects in regard to detection accuracy, processing time. Through experiments, we show requirement of background models in image sequences form a moving camera.

DOI： 10.1109/FCV.2015.7103752

Language：English   Publishing type：Research paper (other academic)  

The segmentation of transparent objects can be very useful in computer vision applications. However, because they borrow texture from their background and have a similar appearance to their surroundings, transparent objects are not handled well by regular image segmentation methods. We propose a method that overcomes these problems using the consistency and distortion properties of a light-field image. Graph-cut optimization is applied for the pixel labeling problem. The light-field linearity is used to estimate the likelihood of a pixel belonging to the transparent object or Lambertian background, and the occlusion detector is used to find the occlusion boundary. We acquire a light field dataset for the transparent object, and use this dataset to evaluate our method. The results demonstrate that the proposed method successfully segments transparent objects from the background.

DOI： 10.1109/ICCV.2015.393

Language：English   Publishing type：Research paper (other academic)  

The purpose of this study is to investigate the relationship between academic achievement and learning patterns of students using e-book logs. Specifically, we examined how students who maintain good academic achievement link among knowledge of different e-books. We hypothesized that good achievers might access e-books sequentially those were used in the same class session and/or consecutive class sessions, for systematically linking among the different knowledge of related e-books. Logs were collected from first-year students in an information science course at Kyushu University. The present study revealed that the good achievers more frequently linked e-books which were used in the same class sessions than the poor achievers. This result suggests that the good achievers more frequently linked knowledge of e-books which deeply related each other.

Language：Japanese   Publishing type：Research paper (scientific journal)  

KIKAN education KADAI-KYOGAKU (interdisciplinary collaborative learning of social issues)

DOI： 10.15017/1495422

Language：English   Publishing type：Research paper (other academic)  

In this paper, we present a meaningful learning environment to visually support e-book learners to effectively construct their knowledge framework. This personalized visualization support is intended to encourage learners to actively locate new knowledge in their own knowledge framework and check the logical consistency of their ideas for clearing up misunderstandings. On the other hand, we also propose to visually support e-book instructors to decide the group distribution for collaborative learning activities based on knowledge structure of learners. To facilitate those visualization supports, we present a method to semi-Automatically construct a course-centered ontology to describe the required information in a map structure.

Language：English   Publishing type：Research paper (other academic)  

This paper presents a method for visualizing students' learning logs using discrete graphs. These logs contain the following four items: attendance, time spent browsing slides, submission of a report and the quiz score for each lesson. The data were collected using learning management systems and the e-text systems. By using these data, we construct graphs for each grade of which the nodes represent all combinations of achievements and failures for the four items. The graphs enable us to observe the features of students' learning activities for each obtained grade. The order in which the above four items are presented changes the visual features of the graph. Moreover, the construction of a graph from the data of the same class held previously enables us to inform students of the learning activities they should avoid. Finally, future research plans regarding this method are presented.

Language：English  

Cyber physical system (CPS) is a general computation concept, in which “Computers (Cyber world)” and “Real world” are integrated via computer networks. In a cyber physical system, there is a loop structure of “observation,” “processing” and “feedback” in the real world: (i) various kinds of data are acquired from our real world using various sensors; (ii) then those data are transferred to computers, or cyber world, and are processed and analyzed; (iii) the analyzed results are fed back to the real world and the real world are modified according to the feedback. Based on this loop structure, the real world is changed, or adjusted. The concept of cyber physical system is well suited for the framework of various IT-based social services, and, in this chapter, we present our research project applying the CPS to social services, especially to an energy management problem, which is one of the most crucial issues for our future society.

DOI： 10.1007/978-3-319-14711-6_17

Language：English   Publishing type：Research paper (other academic)  

Making a reasonable ranking on images in dataset is one of the main objectives for object retrieval challenge, and in this paper we intend to improve the ranking quality. We follow the idea of query expansion in previous researches. Based on the use of bag-of-visual-words model, tf-idf scoring and spatial verification, previous method applied a pointwise style learning in query expansion stage, using but not fully exploring verification results. We intend to extend their learning approach for better discriminative power in retrieval. In re-ranking stage we propose a method using pairwise learning, instead of pointwise learning previously used. We could obtain more reliable ranking on a shortlist of examples. If this verification itself is reliable, a good re-ranking should best preserve this sub-ranking order. Thus in our proposed method, we are motivated to leverage a pairwise learning method to incorporate the ranking sequential information more efficiently. We evaluate and compare our proposed method with previous methods over Oxford 5k dataset, a standard benchmark dataset, where our method achieve better mean average precision and showed better discriminative power.

DOI： 10.1109/FCV.2015.7103703

Language：English   Publishing type：Research paper (other academic)  

Analyses on students' learning behaviors comprise an important thrust in education research. This study focused on e-books system used in the classroom and this system recorded students' learning logs in their daily academic life. These learning logs can be used to analysis students' learning behaviors. By performing partial correlation analysis, the study found that a number of learning behaviors have a significant relation with students' test scores.

Language：English   Publishing type：Research paper (other academic)  

In this paper, we address the problem of image annotation with incomplete labelling, where the multiple objects in each training image are not fully labeled. The conventional one-versus-all SVM (OVA-SVM) that performs fairly well on full labelling decays drastically under the incomplete setting. Recently, structured learning method termed OVA-SSVM is proposed to boost the performance of OVA-SVM by modeling the structured associations of labels and show efficiency under incomplete setting. The OVA-SSVM assumes that each training sample includes a single label and adopts an loss measure of classification style that as long as one of the predicted label is correct, the overall prediction should be considered correct. However, this may not be appropriate for the multi-label annotation task. In this paper, we extend the OVA-SSVM method to the multi-label situation and design a novel image specific structured loss measure to account for the dependencies between predicted labels relying on the image-label associations. Then we develop an efficient optimization algorithm to learn the model parameters. Finally, we present extensive empirical results on two benchmark datasets with various degree of incompletion, and show that proposed method outperforms OVA-SSVM and achieves competitive performance compared with other state-of-the-art methods which are also designed for the issue of incomplete labelling.

DOI： 10.1007/978-3-319-16865-4_46

Language：English   Publishing type：Research paper (other academic)  

Many universities choose the C programming language (C) as the first programming language to teach to students. As novice programmers, students frequently make simple mistakes such as syntax and typographical errors. Students often find it difficult to locate these errors, as students are not yet thoroughly familiar with C's syntax. This situation often causes students to consider programming very dull. It is therefore critical to provide clearer explanation in class, to prevent students losing interest in programming. This study aims to facilitate teaching and learning of C. We propose a system that undergraduate novice programmers may use to locate syntax errors in C. We analyze error types and resolutions using data collected during a programming course, and discuss key findings and their implications for programming education.

Language：English   Publishing type：Research paper (other academic)  

Many universities choose the C programming language (C) as the first programming language to teach to students. As novice programmers, students frequently make simple mistakes such as syntax and typographical errors. Students often find it difficult to locate these errors, as students are not yet thoroughly familiar with C's syntax. Usually educational materials are very useful tools for students to locate errors and find solutions. This study aims to facilitate teaching and learning of C. We propose a system that undergraduate novice programmers may use to easily locate syntax errors in C and get recommendations from educational materials. We analyze error logs of programming and reading logs of educational materials, with the learning by doing mode (learning-practicing-reflection) to discuss key findings and their implications for programming education.

Language：English   Publishing type：Research paper (other academic)  

This paper provides an overview of the Educational Big Data research project at Kyushu University, Japan. This project uses an e-book system called BookLooper. which allows students to browse e-books in Web browser, PC, mobile devices such as smartphone. This paper shows research issues in this project. Currently, about 2,700 first-year students are using the e-book system and approximately 2.2 million log data have been accumulated as of May 20, 2015. This paper describes why we introduce e-book in the University education and initial findings.

Language：English   Publishing type：Research paper (other academic)  

In this paper, we propose a novel method of summarizing lecture slides to enhance preview efficiency and improve students' understanding of the content. Students are often asked to prepare for a class by reading lecture materials. However, this does not always produce good results because the attention span of students is limited. We conducted a survey involving preview of lecture materials by more than 300 students and found that they want summarized materials to preview. Therefore, we developed an automatic summarization method to reduce the original preview materials to a summarized set. Our approach is based on the use of image processing and text processing to extract important pages from lecture materials, and then optimizing the selection of pages in accordance with a specified preview time. We applied the proposed summarization method to lecture slides. In our user study involving more than 300 students, we compared the relative effectiveness of the summarized slides and the original materials in terms of quiz scores, preview achievement ratio, and time spent previewing. We found that students who previewed the summarized slides achieved better scores on pre-lecture quizzes even though they spent less time previewing the material.

Language：English   Publishing type：Research paper (other academic)  

The analysis of learning behavior and identification of learning style from learning logs are expected to benefit instructors and learners. This study describes methods for processing learning logs, such as data collection, integration, and cleansing, developed in Kyushu University. The research aims to analyze learning behavior and identify students' learning style using student's learning logs. Students were clustered into four groups using k-means clustering, and features of their learning behavior were analyzed in detail. We found that Digital Backtrack Learning style is better than Digital Sequential Learning style.

Language：English   Publishing type：Research paper (other academic)  

The purpose of this study is to investigate the relationship between academic achievement and learning patterns of students using e-book logs. Specifically, we examined patterns of students' e-book logs before and after the main content learning in class (that is, 'Preview' and 'Review'). Logs were collected from first-year students in an information science course at Kyushu University. To measure preview and review learning, we analyzed data using three types of measurement: Change indicates how many times a student changed e-books over the course of one hour. Duration indicates how many seconds a student access a given e-book for during one Change (i.e., one turn). Page flip indicates how many pages of a given e-book a student flipped through during one Change. To analyze the relationship between academic achievement and preview/review, the students were categorized into six groups according to their scores on midterm and final (term-end) examinations. For preview, students who had consistent good achievement showed higher values for all three measurements than students who showed poor achievement. In contrast, for review, none of the three measurements showed differences among the six groups. These results suggest that preview is more deeply relevant to academic achievement and assessment than review.

Language：English   Publishing type：Research paper (other academic)  

This paper proposes a method to analyze preview behaviors of students using a learning management system (LMS) and an e-book system. We collected a large number of operation logs from e-books to analyze the process of learning. In addition, we conducted a quiz to test the level of understanding. This study especially focuses on an analysis of the relationship between learning behavior in preview and its effectiveness in the corresponding quiz. We apply a machine learning and classification methodology for behavior analysis. Experimental results report that students who undertake good preview achieve better scores in quizzes.

Language：English   Publishing type：Research paper (scientific journal)  

User-provided image tags are usually incomplete or noisy to describe the visual content of corresponding images. In this paper, we consider defective tagging which covers both incomplete and noisy situations, and address the problem of tag completion where tag assignments of training images are defective. While previous studies on tag completion usually assign equal penalty to empirical loss when processing each missing or noisy tag for each image, we show that this may be suboptimal as the relatedness of each tag to each image varies due to the defective setting. Thus, we introduce an image-tag re-weighting scheme to re-weight the penalty term of each tag to each image considering both image similarities and tag associations, and formulate a unified re-weighted empirical loss function. Experimental evaluations show that embedding proposed re-weighted empirical loss function in state-of-the-art tag completion algorithms achieves significant improvement in dealing with defective tag assignments.

DOI： 10.1109/ICME.2014.6890154

Language：English   Publishing type：Research paper (scientific journal)  

We present a robust background model for object detection and its performance evaluation using the database of the Background Models Challenge (BMC). Background models should detect foreground objects robustly against background changes, such as "illumination changes" and "dynamic changes". In this paper, we propose two types of spatiotemporal background modeling frameworks that can adapt to illumination and dynamic changes in the background. Spatial information can be used to absorb the effects of illumination changes because they affect not only a target pixel but also its neighboring pixels. Additionally, temporal information is useful in handling the dynamic changes, which are observed repeatedly. To establish the spatiotemporal background model, our frameworks model an illumination invariant feature and a similarity of intensity changes among a set of pixels according to statistical models, respectively. Experimental results obtained for the BMC database show that our models can detect foreground objects robustly against background changes.

DOI： 10.1016/j.cviu.2013.10.015

Language：English   Publishing type：Research paper (scientific journal)  

Background modeling and subtraction is an essential task in video surveillance applications. Many researchers have discussed about an improvement of performance of a background model, and a reduction of memory usage or computational cost. To adapt to background changes, a background model has been enhanced by introducing various information including a spatial consistency, a temporal tendency, etc. with a large memory allocation. Meanwhile, an approach to reduce a memory cost cannot provide better accuracy of a background subtraction. To tackle the trade-off problem, this paper proposes a novel framework named "case-based background modeling". The characteristics of the proposed method are (1) a background model is created, or removed when necessary, (2) case-by-case model sharing by some of the pixels, (3) pixel features are divided into two groups, one for model selection and the other for modeling. These approaches realize a low-cost and high accurate background model. The memory usage and the computational cost could be reduced by half of a traditional method and the accuracy was superior to the method.

DOI： 10.1007/s00138-013-0563-4

Language：English  

One of the fundamental problems in computer vision is detecting regions or objects of interest from an image sequence. Background subtraction, which removes a background image from the input image, is widely used for detecting foreground objects in practical applications, since it enables us to detect foreground objects without any previous knowledge of them. However, simple background subtraction often detects not only foreground objects but also a lot of noise regions, because it is quite sensitive to background changes. In general, background changes which occur in outdoor scenes can be mainly classified into two types: • Illumination changes – changes caused by lighting conditions such as the sun rising, setting, or being blocked by clouds, • Dynamic changes – changes caused by the swaying motion of tree branches, leaves and grass, fleeting cloud, waves on water and so on.

DOI： 10.1201/b17223

Language：English   Publishing type：Research paper (other academic)  

Nowadays, it is a hot topic to analyze the huge amount of data in the world. This issue also exists in the learning during students' life. The learning data are collected only to record students' learning status. As a result, most learning data are not used to improve the quality of learning for students. In this paper, we propose an order made education system, which can recommend students to select the courses they want to learn. In order to analyze students' learning behaviors, we collect students' learning data by using mobile devices.

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we propose a large-scale image annotation system for the ImageCLEF 2014 Scalable Concept Image Annotation task. The annotation task, of this year, concentrated on developing annotation algorithms that rely only on data obtained automatically from the web. Since the sophisticated SVM based annotation techniques had been widely applied in the task last year (ImageCLEF 2013), for the task this year, we also adopt the SVM based annotation techniques and put our effort mainly on obtaining more accurate concepts assignment for training images. More specifically, we proposed a two-fold scheme to assign concepts to unlabeled training images: (1) A traditional process which stems the extracted web data of each training image from textual aspect, and make concepts assignment based on the appearance of each concept. (2) An additional process which leverages the deep convolutional network toolbox Overfeat to predict labels (in ImageNet nouns) for each training image from visual aspect, then the predicted tags are mapped to concepts in ImageCLEF based on WordNet synonyms and hyponyms with semantic relations. Finally, the allocated concepts for each training image are generated based on a fusion step of the two-fold concepts assignment processes. Experimental results show that the proposed concepts assignment scheme is efficient to improve the assignment results of traditional textual processing and to allocate reasonable concepts for training images. Consequently, with an efficient SVMs solver based on S-tochastic Gradient Descent, our annotation systems achieves competitive performance in the annotation task.

Language：English   Publishing type：Research paper (other academic)  

This paper presents an incremental learning method for hand gesture recognition that learns the individual movements in each gesture of a user. To recognize the movement, we use a subunit-based dynamic time warping method, which treats a hand movement as a sequence of ubmovements. In our method, each hand movement is decomposed into submovements and the arrangement of submovements is reflected in the training sample database. Experimental results from the lassification of ten gestures demonstrate that our method can improve the recognition rate compared with a method without incremental learning. In addition, the experimental results show that incremental learning of a single class of gestures can improve the recognition rate of multi-class gestures using our method.

DOI： 10.1007/978-3-319-11755-3_7

Language：English   Publishing type：Research paper (other academic)  

A subunit-based Dynamic Time Warping (DTW) approach is proposed for hand movement recognition. Two major contributions distinguish the proposed approach from conventional DTW. (1) A set of hand movement subunits is constructed using a data-driven method. The common sub-movements (subunits) are shared across hand gestures to obtain a smaller training data size and search space to improve recognition performance. (2) A similarity measure robust to variability is offered using subunit-to-subunit matching to absorb the difference between two similar sub-sequences belonging to the same subunit, and only keeping the distances between sub-sequences that relate to different subunits. Our experimental results demonstrate the efficiency and accuracy of the proposed approach.

DOI： 10.1007/978-3-642-41181-6_68

Language：English   Publishing type：Research paper (other academic)  

For the task of image annotation, traditional probabilistic topic models based on Latent Dirichlet Allocation (LDA) [1], assume that an image is a mixture of latent topics. An inevitable limitation of LDA is the inability to model topic correlation since topic proportions of an image are generated independently. Motivated by Correlated Topic Model (CTM) [2] which derives from natural language processing to model topic correlation of a document, we extend the popular LDA based models (corrLDA [3], sLDA-bin [4], trmmLDA [5]) to CTM based models (corrCTM, sCTM-bin, trmmCTM). We present a comprehensive comparison between CTM based and LDA based models on three benchmark datasets, illustrating the superior annotation performance of proposed CTM based models, by means of propagating topic correlation among image features and annotation words.

DOI： 10.1109/ICME.2013.6607531

Language：English   Publishing type：Research paper (other academic)  

Recently, weighted k nearest neighbor based label prediction model combined with distance metric learning (KNN+ML) [10,14,17], has become more attractive and showed exciting results on image annotation task. Usually, in KNN+ML framework, a uniform distance metric is learned given a collection of similar/dissimilar image pairs from training data. Thus, for a couple of images, their distance is globally unique. However, this might not be sufficient for label prediction on annotation task because it is impossible to distinguish the multiple labels attached to each image. In this paper, we are motivated to learn multiple label-specific distance metrics, and measure the distance of an image pair under different labels' distance metrics. We also propose a novel label specific prediction model, in which the weight of each label is determined by its specific distance value rather than previous global distance value. Compared with previous KNN+ML methods, our proposed method is able to exactly discriminate each label in each neighbor, and efficiently reduce the prediction of false positive and false negative labels. Extensive experimental results on three benchmark datasets demonstrate that proposed method achieves more accurate annotation results and competitive overall performance.

DOI： 10.1007/978-3-642-41181-6_11

Language：English   Publishing type：Research paper (other academic)  

This paper introduces a new open dataset "Actions for Cooking Eggs (ACE) Dataset" and summarizes results of the contest on "Kitchen Scene Context based Gesture Recognition", in conjunction with ICPR2012. The dataset consists of naturally performed actions in a kitchen environment. Five kinds of cooking menus were actually performed by five different actors, and the cooking actions were recorded by a Kinect Sensor. Color image sequences and depth image sequences are both available. Besides, action label was given to each frame. To estimate the action label, action recognition method has to analyze not only actor's action, but also scene contexts such as ingredients and cooking utensils. We compare the submitted algorithms and the results in this paper.

DOI： 10.1007/978-3-642-40303-3_18

Language：English   Publishing type：Research paper (scientific journal)  

This paper discusses about object detection based on spatio-temporal light field sensing. Our proposed method generates an arbitrary in-focus plane in the surveillance scene, and the background region can be filtered out by out-focusing. A new feature representation, called Local Ray Pattern (LRP), is introduced to evaluate the spatial consistency of light rays. The combination of LRP and GMM-based background modeling realizes object detection on the in-focus plane. Experimental results demonstrate the effectiveness and applicability for video surveillance.

DOI： 10.2197/ipsjtcva.5.129

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we propose an efficient algorithm to recognize actions of human interaction. Unlike previous algorithms using two participants' actions, the proposed algorithm estimates the action contribution of participants to determine which participant's action is the major action for correct interaction recognition. To estimate this contribution, we construct a contribution interaction model for each interaction category in which both participants carry out major actions. Using the contribution models, we design a method that automatically estimates the contribution of participants and classifies interaction samples into "co-contribution" and "single-contribution" interactions. At the same time, the major actions in the "single-contribution" interactions are determined. We evaluate our method on the UT-interaction dataset and our original interaction dataset (LIMU). Recognition results indicate the robustness of the proposed method and the high estimation accuracy obtained: estimation accuracies of 96 and 98% in set 1 and set 2 of the UT dataset, respectively, and 97.8% in the LIMU dataset. Based on the estimation results, we extract the major action information for interaction recognition. Average recognition accuracies of 93.3% in set 1 and 91.7% in set 2 of the UT dataset were obtained. Our result is at least 5% better than those obtained with previous algorithms. For the LIMU dataset, recognition accuracy reached 91.1%. It was 8.9% higher than the recognition result without contribution estimation.

DOI： 10.1002/tee.21850

Language：English   Publishing type：Research paper (other academic)  

We propose an intuitive operation based information retrieval system 'Clickable Real World (CRW)'. If a user which uses this system takes a picture of a landmark in the world, some related information is displayed on a smartphone. One of the key research issues is how to estimate appropriate keywords in order to retrieve information related to the target. Our strategy utilizes a lot of training samples, which consist of images, tags and geolocation where the image was taken, shared on the Web (such as flickr and Picasa). Then, a keyword table is created based on the consistency of geolocation and visual feature. We developed a prototype version of CRW on a smartphone, and conducted a field experiment in Kyoto city, Japan.

DOI： 10.1109/FCV.2013.6485453

Language：English   Publishing type：Research paper (other academic)  

A man-machine interface plays an important role to convey an intention from a user to machine. Nowadays, vision-based solution has been attracting a lot of attention since it does not require any attachment sensors on the body. One of the most famous applications is to control TV operations by hand gestures. Instead of bothersome operations using a specific controller, a user can send a command to a system by intuitive hand gestures. Most of previous studies have been focused on a strategy to detect/recognize hand motions, and the design concept to realize high usability of the application got less attention. This paper discusses how to design a user-friendly and also machine-friendly hand gesture application. Our concept gives a user an opportunity to customize a TV control interface by selecting hand shapes and hand motions through interaction with a hand gesture selection system.

DOI： 10.1109/FCV.2013.6485473

Language：English   Publishing type：Research paper (other academic)  

For the task of image annotation, traditional methods based on probabilistic topic model, such as correspondence Latent Dirichlet Allocation (corrLDA) [1], assumes that image is a mixture of latent topics. However, this kind of models is unable to directly model correlation between topics since topic proportions of an image are generated independently. Our model, called correspondence Correlated Topic Model (corrCTM), extends Correlated Topic Model (CTM) [2] from natural language processing to capture topic correlation from covariance structure of more flexible model distribution. Unlike previous LDA based models, topic proportions are correlated with each other in proposed corrCTM. And the topic correlation propagates from image features to annotation words through a generative process, and finally correspondence between images and words could be generated. We derive an approximate inference and estimation algorithm based on variational method. We examine the performance of our model on two benchmark image datasets, show improved performance over corrLDA for both annotation and modeling word correlation.

DOI： 10.1109/FCV.2013.6485488

Language：English   Publishing type：Research paper (other academic)  

Object detection is an important task for computer vision applications. Many researchers have proposed a lot of methods to detect the objects through the background modeling. Most of previous approaches model the background independently for each pixel and detect foreground objects based on it. Then, it is difficult for the background model to deal with illumination changes, which cause significant intensity changes as in the case that a foreground object appears. To solve this problem, in this paper, we propose a new background model considering the similarity in the intensity changes among pixels. In particular, we classify all the pixels into several clusters based on the similarity of their intensity changes. Then, focusing on each cluster, we can easily identify whether the significant intensity changes are caused by foreground objects or illumination changes. This is because, if the illumination changes, most of the pixels belonging to the same cluster exhibit the similar intensity changes.

DOI： 10.1109/FCV.2013.6485504

Language：English   Publishing type：Research paper (other academic)  

We present a robust background model for object detection and report its evaluation results using the database of Background Models Challenge (BMC). Our background model is based on a statistical local feature. In particular, we use an illumination invariant local feature and describe its distribution by using a statistical framework. Thanks to the effectiveness of the local feature and the statistical framework, our method can adapt to both illumination and dynamic background changes. Experimental results, which are done thanks to the database of BMC, show that our method can detect foreground objects robustly against background changes.

DOI： 10.1007/978-3-642-37410-4_30

Language：English   Publishing type：Research paper (scientific journal)  

In this paper, we propose a new method to calculate local features. We extend the FAST corner detector to the spatiotemporal space to extract the shape and motion information of human actions. And a compact peak-kept histogram of oriented spatiotemporal gradients (CHOG3D) is proposed to calculate local features. CHOG3D is calculated in a small support region of a feature point, and it employs the simplest gradient, the first-order gradient, for descriptor calculation. Through parameter training, the proper length of the CHOG3D is determined to be 80 elements, which is 1/12.5 times the dimension of HOG3D in the KTH dataset. In addition, it keeps the peak value of quantized gradient to represent human actions more exactly and distinguish them more correctly. CHOG3D overcomes the disadvantages of the complex calculation and huge length of the descriptor HOG3D. From a comparison of the computation cost, CHOG3D is 7.56 times faster than HOG3D in the KTH dataset. We apply the algorithm for human action recognition with support vector machine. The results show that our method outperforms HOG3D and some other currently used algorithms.

DOI： 10.1002/tee.21793

Language：English   Publishing type：Research paper (scientific journal)  

Recently, office buildings and commercial facilities are getting larger, and emergency evacuation guidance procedures are urgently required. To support evacuation planning, several kinds of evacuation simulations have been proposed. These use walking velocity models that were generated depending on actual pedestrian flow to define an agent's velocity. However, most of these conventional models have been simplified and it is difficult to reproduce complex evacuation scenarios faithfully. In this paper, we propose a walking velocity model for accurate pedestrian simulations. The model presents the relation between pedestrian density and velocity distribution; it was generated through analyzing flows observed from actual evacuation drills. We modeled dense pedestrian flows using the flow data with conventional models to improve simulation performance. In addition, we introduced a method of representing difference among individuals. The validity of the model is confirmed by experimenting with the pedestrian simulator.

DOI： 10.1541/ieejeiss.133.1779

Language：English   Publishing type：Research paper (scientific journal)  

Current object-recognition algorithms use local features, such as scale-invariant feature transform (SIFT) and speeded-up robust features (SURF), for visually learning to recognize objects. These approaches though cannot apply to transparent objects made of glass or plastic, as such objects take on the visual features of background objects, and the appearance of such objects dramatically varies with changes in scene background. Indeed, in transmitting light, transparent objects have the unique characteristic of distorting the background by refraction. In this paper, we use a single-shot light field image as an input and model the distortion of the light field caused by the refractive property of a transparent object. We propose a new feature, called the light field distortion (LFD) feature, for identifying a transparent object. The proposal incorporates this LFD feature into the bag-of-features approach for recognizing transparent objects. We evaluated its performance in laboratory and real settings.

DOI： 10.1109/CVPR.2013.359

Language：English   Publishing type：Research paper (scientific journal)  

Clickable Real World is a new framework to realize an intuitive information search with a mobile terminal. To achieve the goal, we tackle two challenging tasks. One is landmark detection from an observing scene. Our approach detects a landmark based on an image prior. The prior is not given manually. Instead, it is generated automatically from the training samples collected from photo sharing website. Another challenging task is image annotation assisted by geolocation. We use the location of a user who uses a mobile terminal, and geolocation where the training sample images were taken. Two probabilistic models are generated to achieve image annotation. One is image-based labeling which utilizes the co-occurrence between image features and label features. The other is label-based localization which uses the consensus about the label given around the geolocation among photographers. We combine two probabilistic approaches to improve the accuracy of image annotation. We demonstrate this approach for 87 scenes in the world.

DOI： 10.1541/ieejeiss.133.142

Language：English   Publishing type：Research paper (scientific journal)  

Background modeling and subtraction is an essential task in video surveillance applications. Most traditional studies use information observed in past frames to create and update a background model. To adapt to background changes, the background model has been enhanced by introducing various forms of information including spatial consistency and temporal tendency. In this paper, we propose a new framework that leverages information from a future period. Our proposed approach realizes a low-cost and highly accurate background model. The proposed framework is called bidirectional background modeling, and performs background subtraction based on bidirectional analysis, i.e., analysis from past to present and analysis from future to present. Although a result will be output with some delay because information is taken from a future period, our proposed approach improves the accuracy by about 30% if only a 33-millisecond of delay is acceptable. Furthermore, the memory cost can be reduced by about 65% relative to typical background modeling.

DOI： 10.1109/CVPR.2013.258

Language：English   Publishing type：Research paper (scientific journal)  

A method is described for analyzing the image disparity of 3D video. This method was used to analyze the image disparity charactristics of forty 3D movies. The percentage of image disparities that were out-of-range by one degree or more was higher in the convergent direction than in the divergent direction. Characteristics of the image disparity for 3D video are discussed in relation to the results.

DOI： 10.3169/itej.67.J400

Language：English   Publishing type：Research paper (scientific journal)  

In these days, "early recognition" of gesture patterns has been studied by many researchers. Early recognition is a method to make a decision of gesture recognition at the beginning part of it. In traditional method, the key postures for a gesture are utilized for recognition and early recognition is performed frame-by-frame. However, this method has a problem that computational time in recognition processing increases in proportion to size of posture database. If the processing time becomes longer, some input frames will be ignored from the processing. It results in lower recognition accuracy. In this paper, we introduce a hash-based approach to search the posture database. It realizes real-time processing, and keep high performance of recognition.

DOI： 10.1007/s10015-012-0085-6

Language：English   Publishing type：Research paper (other academic)  

We report evaluation results of an integrated background modeling based on spatio-temporal features. The background modeling method consists of three complementary approaches: pixel-level background modeling, region-level one and frame-level one. The pixel-level background model uses the probability density function to approximate background model. The PDF is estimated non-parametrically by using Parzen density estimation. The region-level model is based on the evaluation of the local texture around each pixel while reducing the effects of variations in lighting. The frame-level model detects sudden, global changes of the the image brightness and estimates a present background image from input image referring to a background model image. Then, objects are extracted by background subtraction. Fusing these approaches realizes robust object detection under varying illumination.

DOI： 10.1109/CVPRW.2012.6238920

Language：English   Publishing type：Research paper (other academic)  

This paper discusses hand shapes for Human Computer Interface. Usually, a hand gesture based Human Computer Interface is developed by human centered design concept. A system designer or developer tends to select hand shapes by himself/herself without verifying practical effectiveness from the standpoint of system aspect. Instead, a methodology of training and recognition of hand shapes is often discussed. On the other hand, this paper listens to system's voice; which hand shape is easy to be recognized, which is easy to be confused and so on. Actually, 37 kinds of tentative hand shapes were investigated from the viewpoint of system-friendly hand shape. Based on the result, a supporting system was developed for a system designer, which helps to find appropriate hand shapes which satisfy both "user-friendly" and "system- friendly" demand.

Language：English   Publishing type：Research paper (scientific journal)  

Human action recognition is applied in a wide field, such as video surveillance, intelligent interface, and intelligent robots. However, since various action classes, complex surrounding, interaction with objects, et al., it is still a complex problem to be solved. In this paper, we propose a method combining the Self-Organizing Map(SOM) and the classifier k-Nearest Neighbor algorithm (k-NN) to recognize human actions. We represent human actions in the form of local features using a compact descriptor, a histogram of oriented gradient in spatio-temporal 3D space(CHOG3D), which was proposed by us in the paper 1). Then we adopt SOM for feature training to extract key features of action information. With these key features, we adopt k-NN for action recognition. In our experiments, we test the optimal map size of SOM and the proper value k of k-NN for correct recognition. Our method is tested on KTH, Weizmann and UCF datasets, and results certify its efficiency.

Language：English   Publishing type：Research paper (other academic)  

In this paper, we propose a method combining visual local features and depth image information to recognize cooking gestures. We employ the feature calculation method [2] which used extended FAST detector and a compact descriptor CHOG3D to calculate visual local features. We pack the local features by BoW in frame sequences to represent the cooking gestures. In addition, the depth images of hands gestures are extracted and integrated spatio-temporally to represent the position and trajectory information of cooking gestures. The two kinds of features are used to describe cooking gestures, and recognition is realized by employing the SVM. In our method, we determine the gesture class for each frame in cooking sequences. By analyzing the results of frames, we recognize cooking gestures in a continue frame sequences of cooking menus, and find the temporal positions of the recognized gestures.

DOI： 10.1145/2390776.2390785

Language：English   Publishing type：Research paper (other academic)  

Estimation of electric power consumption of individuals based on human action analysis is presented. It is a key tool to reduce the energy consumption.

DOI： 10.1109/ICCPS.2012.29

Language：English   Publishing type：Research paper (other academic)  

We propose a new approach for early gesture recognition. Early gesture recognition is a method to recognize sequential posture patterns at their beginning parts. Using early gesture recognition, we can reduce the delay and increase the interactivity of the system. The key issue of early recognition problem is how to recognize the beginning part of gesture. Generally, a reference gesture consists of a sequence of postures. Therefore, to realize early recognition we have to select beginning parts of the posture sequences. In this paper, we propose a method to obtain such beginning parts of the posture sequences, and demonstrate its effectiveness through some experiments.

DOI： 10.1109/FCV.2011.5739719

Language：English   Publishing type：Research paper (scientific journal)  

Object detection is an important task for computer vision applications. Many researchers have proposed a number of methods to detect the objects through background modeling. To adapt to "illumination changes" in the background, local feature-based background models are proposed. They assume that local features are not affected by background changes. However, "motion changes", such as the movement of trees, affect the local features in the background significantly. Therefore, it is difficult for local feature-based models to handle motion changes in the background. To solve this problem, we propose a new background model in this paper by applying a statistical framework to a local feature-based approach. Our proposed method combines the concepts of statistical and local feature-based approaches into a single framework. In particular, we use illumination invariant local features and describe their distribution by Gaussian Mixture Models (GMMs). The local feature has the ability to tolerate the effects of "illumination changes", and the GMM can learn the variety of "motion changes". As a result, this method can handle both background changes. Some experimental results show that the proposed method can detect the foreground objects robustly against both illumination changes and motion changes in the background.

DOI： 10.2197/ipsjtcva.3.198

Language：English   Publishing type：Research paper (scientific journal)  

An adaptive background model plays an important role for object detection in a scene which includes illumination changes. An updating process of the background model is utilized to improve the robustness against illumination changes. However, the process sometimes causes a false-negative problem when a moving object stops in an observed scene. A paused object will be gradually trained as the background since the observed pixel value is directly used for the model update. In addition, the original background model hidden by the paused object cannot be updated. If the illumination changes behind the paused object, a false-positive problem will be caused when the object restarts to move. In this paper, we propose 1) a method to inhibit background training to avoid the falsenegative problem, and 2) a method to update an original background region occluded by a paused object to avoid the false-positive problem. We have used a probabilistic approach and a predictive approach of the background model to solve these problems. The great contribution of this paper is that we can keep paused objects from being trained by modeling the original background hidden by them. And also, our approach has an ability to adapt to various illumination changes. Our experimental results show that the proposed method can detect stopped objects robustly, and in addition, it is also robust for illumination changes and as efficient as the state-of-the-art method.

DOI： 10.2197/ipsjtcva.3.148

Language：English   Publishing type：Research paper (other academic)  

The growth of photo-sharing website such as Flickr and Picasa enables us to access the billions of images easily. Recent years, many researchers leverage such photo-sharing site to tackle the image annotation problem. The aim of the image annotation is to give a proper label to an unknown image. Generally, image features and label features are used to acquire the relationship between them. Meanwhile, we use not only such image and label features but also geolocation which indicate the information where the image was taken. We formulate the image annotation problem as two important issues; image-based labeling and label-based localization. The former issue is to estimate a proper label from a given image. The latter is the issue to estimate the location from the label. Our approach combine these two estimation strategies. We conducted some experiments and found that our approach outperformed the traditional approach.

DOI： 10.1109/ACPR.2011.6166619

Language：English   Publishing type：Research paper (other academic)  

Background modeling has been widely researched to detect moving objects from image sequences. Most approaches have a falsenegative problem caused by a stopped object. When a moving object stops in an observing scene, it will be gradually trained as background since the observed pixel value is directly used for updating the background model. In this paper, we propose 1) a method to inhibit background training, and 2) a method to update an original background region occluded by stopped object. We have used probabilistic approach and predictive approach of background model to solve these problems. The great contribution of this paper is that we can keep paused objects from being trained.

DOI： 10.1007/978-3-642-22822-3_2

Language：English   Publishing type：Research paper (scientific journal)  

We propose an approach to achieving early recognition of gesture patterns. Early recognition is a method for recognizing sequential patterns at their earliest stage. Therefore, in the case of gesture recognition, we can get a recognition result for human gestures before the gestures are finished. The most difficult problem in early recognition is knowing when the system has determined the result. Most traditional approaches suffer from this problem, since gestures are often ambiguous. At the start of a gesture, in particular, it is very difficult to determinate the recognition result since insufficient input data have been observed. Therefore, we have improved on the traditional approach by using a self-organizing map.

DOI： 10.1007/s10015-011-0917-9

Language：English   Publishing type：Research paper (other academic)  

We propose a new method of background modeling for object detection. Many background models have been previously proposed, and they are divided into two types: "pixel-based models" which model stochastic changes in the value of each pixel and "spatial-based models" which model a local texture around each pixel. Pixel-based models are effective for periodic changes of pixel values, but they cannot deal with sudden illumination changes. On the contrary, spatial-based models are effective for sudden illumination changes, but they cannot deal with periodic change of pixel values, which often vary the textures. To solve these problems, we propose a new probabilistic background model integrating pixel-based and spatial-based models by considering the illumination fluctuation in localized regions. Several experiments show the effectiveness of our approach.

DOI： 10.1007/978-3-642-19282-1_18

Language：English   Publishing type：Research paper (other academic)  

We propose an approach to achieve early recognition of gesture patterns. Early recognition is a method to recognize sequential patterns at their beginning parts. Therefore, in the case of gesture recognition, we can get a recognition result of human gestures before the gestures have finished. The most difficult problem of early recognition is that when the system determines the recognition result. Most traditional approaches suffer from this problem since the gestures comprehend ambiguity. Especially at the beginning part of them, it is very difficult to determinate the recognition result since enough input data has not been observed yet. Therefore, we have improved traditional approach by using Self-Organizing Map.

Language：English   Publishing type：Research paper (other academic)  

In this paper, a new action recognition system is proposed, which employs 3D FAST corner detection in ROI, compact 3D descriptor to represent action information, and SOM to learn and recognize actions. Through detecting 3D FAST corners in ROI, action information of shape and motion can be obtained, and noise corners can be deleted at the same time. Furthermore, based on 3D HOG, we produce a simpler descriptor which is proposed by shortening the support region of interest points, combining symmetric bins after orientation quantization using icosahedron, and keeping the top value bin of quantized histogram. Compared with the descriptor before adjustment, our descriptor uses only 80 bins other than 960 bins to describe one interest point, which saves much computation time and memory. Our frame matching experiment on descriptor also certifies that our descriptor outperforms the previous one. Our descriptor is applied to recognize actions on KTH and Hollywood databases, and the results show that it performs well.

DOI： 10.1109/TENCON.2010.5686694

Language：English   Publishing type：Research paper (other academic)  

We propose a sophisticated method for background modeling based on spatio-temporal features. It consists of three complementary approaches: pixel-level background modeling, region-level one and frame-level one. The pixel-level background model uses the probability density function to approximate background model. The PDF is estimated non-parametrically by using Parzen density estimation. The region-level model is based on the evaluation of the local texture around each pixel while reducing the effects of variations in lighting. The frame-level model detects sudden, global changes of the the image brightness and estimates a present background image from input image referring to a background model image. Then, objects are extracted by background subtraction. Fusing their approaches realizes robust object detection under varying illumination, which is shown in several experiments.

DOI： 10.1007/978-3-642-12307-8_19

Language：English   Publishing type：Research paper (other academic)  

In the Sensing Web[1], a variety of sensors are installed dispersively, and, from those sensors, we acquire various information of the real world events. Although we can acquire a certain kind of information from each of the sensors separately, such information is fragmentary and integration or structurization of sensory data captured by multiple sensors is quite important for us to acquire truly meaningful information of the real world. From this point of view, we have researched into organization and presentation of distributed sensory data in the Sensing Web project. In this paper, we will present our research activity, especially wide-area object tracking, and some of demonstrative experiments.

DOI： 10.1007/978-3-642-14058-7_66

Language：English   Publishing type：Research paper (scientific journal)  

We propose a new method for background modeling based on combination of multiple models. Our method consists of three complementary approaches. The first one, or the pixel-level background modeling, uses the probability density function to approximate background model, where the PDF is estimated non-parametrically by using Parzen density estimation. Then the pixel-level background modeling can adapt periodical changes of pixel values. The regionlevel background modeling is based on the evaluation of local texture around each pixel, which can reduce the effects of variations in lighting. It can adapt gradual change of pixel value. The frame-level background modeling detects sudden and global changes of the image brightness and estimates a present background image from input image referring to a model background image, and foreground objects can be extracted by background subtraction. In our proposed method, integrating these approaches realizes robust object detection under varying illumination, whose effectiveness is shown in several experiments.

DOI： 10.2197/ipsjtcva.2.156

Language：English   Publishing type：Research paper (other academic)  

In this paper, an action recognition system was invented by proposing a compact 3D descriptor to represent action information, and employing self-organizing map (SOM) to learn and recognize actions. Histogram Of Gradient 3D (HOG3D) performed better among currently used descriptors for action recognition. However, the calculation of the descriptor is quite complex. Furthermore, it used a vector with 960 elements to describe one interest point. Therefore, we proposed a compact descriptor, which shortened the support region of interest points, combined symmetric bins after orientation quantization. In addition, the top value bin of quantized vector was kept instead of setting threshold experimentally. Comparing with HOG3D, our descriptor used 80 bins to describe a point, which reduced much computation complexity. The compact descriptor was used to learn and recognize actions considering the probability of local features in SOM, and the results showed that our system outperformed others both on KTH and Hollywood datasets.

DOI： 10.1007/978-3-642-17534-3_48

Language：English   Publishing type：Research paper (other academic)  

We propose an approach to achieve early recognition of gesture patterns. We assume that there are two people who interact with a machine, a robot or something. In such a situation, a gesture of a person often has a relationship with a gesture of another person. We exploit such a relationship to realize early recognition of gesture patterns. Early recognition is a method to recognize sequential patterns at their beginning parts. Therefore, in the case of gesture recognition, we can get a recognition result of human gestures before the gestures have finished. Recent years, some approaches have been proposed. In this paper, we expand the application range of early recognition to multiple people based on the co-occurrence of gesture patterns. In our approach, we use Self-Organizing Map to represent gesture patterns of each person, and associative memory based approach learns the relationship between co-occurring gestures. In the experiments, we have found that our proposed method achieved the early recognition more accurately and earlier than the traditional approach.

DOI： 10.1007/978-3-642-17534-3_53

Language：English   Publishing type：Research paper (other academic)  

"Clickable Real World" is a new methodology to retrieve real-time real-world information from the web. The key point is that the query here is the name and the attributes of a given landmark, which is given by taking a picture of the landmark by a mobile phone camera. The user can feel to directly click the landmark in front of him/her by the shutter clicks. One of the great advantages of our approach is to use open image database such as Flickr, Picasa, or so on, to identify a landmark. Images on such open databases are automatically updated and some proper keywords are given by photographers in the world. Therefore, we need not prepare reference images which are required to identify a landmark. In addition, a matched image might have proper labels which indicate the landmark. In this paper, we will explain the overview of Clickable Real World and report some experimental results.

DOI： 10.1109/TENCON.2010.5686550

Language：English   Publishing type：Research paper (other academic)  

Gabor-based face representation has achieved enormous success in face recognition. However, one drawback of Gabor-based face representation is the huge amount of data that must be stored. Due to the nonlinear structure of the data obtained from Gabor response, classical linear projection methods like principal component analysis fail to learn the distribution of the data. A nonlinear projection method based on a set of self-organizing maps is employed to capture this nonlinearity and to represent face in a new reduced feature space. The Multiple Self-Organized Gabor Features (MSOGF) algorithm is used to represent the input image using all winner indices from each SOM map. A new local matching algorithm based on the similarity between local features is also proposed to classify unlabeled data. Experimental results on FERET database prove that the proposed method is robust to expression variations.

DOI： 10.1109/ICPR.2010.713

Language：Japanese   Publishing type：Research paper (scientific journal)  

H-029 Construction of agricultural work recognition system using wearable sensors

Language：English   Publishing type：Research paper (scientific journal)  

We propose a system for counting the number of pedestrians in real-time. This system estimates "how many pedestrians are and where they are in video sequences" by the following procedures. First, candidate regions are segmented into blobs according to background subtraction. Second, a set of features are extracted from each blob and a neural network estimates the number of pedestrians corresponding to each set of features. To realize real-time processing, we used only simple and valid features, and the adaptive background modeling using Parzen density estimation, which realizes fast and accurate object detection in input images. We also validate the effectiveness of the proposed system by several experiments.

DOI： 10.1016/j.sbspro.2010.01.028

Language：English   Publishing type：Research paper (scientific journal)  

We propose a new method for background modeling. Our method is based on the two complementary approaches. One uses the probability density function (PDF) to approximatebackground model. The PDF is estimated non-parametrically by using Parzen density estimation. Then, foreground object is detected based on the estimated PDF. The method is based on the evaluation of the local texture at pixel-level resolution which reduces the effects of variations in lighting. Fusing those approachs realizes robust object detection under varying illumination. Several experiments show the effectiveness of our approach.

DOI： 10.2201/NiiPi.2010.7.4

Language：English   Publishing type：Research paper (scientific journal)  

Background modeling has been widely researched to detect moving objects from image sequences. It is necessary to adapt the background model various changes of illumination condition. Recent years, a hybrid type of background model which consists of more than one background model has been used for object detection since it is very adaptable to illumination changes. In this paper, we also propose a new hybrid type of background model named "Hybrid Spatial-Temporal Background Model". Our model consists of two different kinds of background models. One is pixel-level background model which adapts to long-term illumination changes. The other is spatial-temporal background model which adapts to short-term illumination changes. Our experimental results demonstrate superiority of our method to some related works.

DOI： 10.1541/ieejeiss.130.1524

Language：English   Publishing type：Research paper (other academic)  

Background modeling has been widely researched to detect moving objects from image sequences. It is necessary to adapt the background model various changes of illumination condition. Recent years, a hybrid type of background model which consists of more than one background model has been used for object detection since it is very robust for illumination changes. In this paper, we also propose a new hybrid type of background model named "Hybrid Spatial-Temporal Background Model". Our model consists of two different kinds of background models. One is pixel-level background model which is robust for long-term illumination changes. The other is spatial-temporal background model which is robust for short-term illumination changes. Our experimental results demonstrate superiority of our method to some related works.

DOI： 10.1109/AVSS.2009.12

Language：English   Publishing type：Research paper (other academic)  

Vision-based motion capture is getting popular for acquiring human motion information in various interactive applications. To enlarge its applicability, we have been developing a vision-based motion capture system which can estimate the postures of multiple people simultaneously using multiview image analysis. Our approach is divided into the following two phases: at first, extraction, or segmentation, of each person in input multiview images; then, posture analysis for one person is applied to the segmented region of each person. The segmentation is realized in the voxel space, which is reconstructed by visual cone intersection of multiview silhouettes. Here, a graph cut algorithm is employed to achieve optimal segmentation. Posture analysis is based on a model-based approach, where a skeleton model of human figure is matched with the multiview silhouettes based on a particle filter and physical constraints on human body movement. Several experimental studies show that the proposed method acquires human postures of multiple people correctly and efficiently even when they touch each otter.

DOI： 10.1007/978-3-642-04146-4_49

Language：English   Publishing type：Research paper (other academic)  

We propose a new gesture recognition method which is called "early recognition". Early recognition is a method to recognize sequential patterns at their beginning parts. Therefore, in the case of gesture recognition, we can get a recognition result of human gestures before the gestures have finished. We realize early recognition by using sparse codes of Self-Organizing Map.

DOI： 10.1007/978-3-642-02397-2_14

Language：English   Publishing type：Research paper (other academic)  

We propose a method to eliminate unnecessary neurons in Variable-Density Self-Organizing Map. We have defined an energy function which denotes the error of the map, and optimize the energy function by using graph cut algorithm. We conducted experiments to investigate the effectiveness of our approach.

DOI： 10.1007/978-3-642-02397-2_30

Language：English   Publishing type：Research paper (scientific journal)  

Purpose - The purpose of this paper is to present an approach to improve the accuracy of estimating feature points of human body on a vision-based motion capture system (MCS) by using the variable-density self-organizing map (VDSOM). Design/methodology/approach - The VDSOM is a kind of self-organizing map (SOM) and has an ability to learn training samples incrementally. The authors let VDSOM learn 3D feature points of human body when the MCS succeeded in estimating them correctly. On the other hand, one or more 3D feature point could not be estimated correctly, the VDSOM is used for the other purpose. The SOM including VDSOM has an ability to recall a part of weight vector which have learned in the learning process. This ability is used to recall correct patterns and complement such incorrect feature points by replacing such incorrect feature points with them. Findings - Experimental results show that the approach is effective for estimation of human posture robustly compared with the other methods. Originality/value - The proposed approach is interesting for the collaboration between an MCS and an incremental learning.

DOI： 10.1108/17563780910959910

Language：English   Publishing type：Research paper (scientific journal)  

We propose a new method for background modeling. Our method is based on the two complementary approaches. One uses the probability density function(PDF) to approximate background model. The PDF is estimated non-parametrically by using Parzen density estimation. And foreground object is detected based on the estimated PDF. The other method is based on the evaluation of the local texture at pixel-level resolution while reducing the effects of variations in lighting. Fusing their approach realize robust object detection under varying illumination. Several experiments show the effectiveness of our approach.

DOI： 10.1007/978-3-540-92957-4_56

Language：English   Publishing type：Research paper (scientific journal)  

We propose a new method to create adaptive background models. Traditionally, each pixel has an adaptive background model which consists of Gaussian mixtures. Each model can approximate small changes and periodic changes of pixel values and it helps us to detect moving objects. However, it cannot adapt to some illumination changes such as gradually varying illumination, precipitously varying illumination and so on. The other model such as using a texture or using prediction of pixel value is effective to handle these changes. Therefore, a hybrid background model which is combined with more than two kind of models. In our approach, we use two different types of the background model. The one is the stochastic background model. The other is the predictive background model based on the exponential smoothing.

DOI： 10.1541/ieejeiss.129.846

Language：English   Publishing type：Research paper (other academic)  

The purpose of our work is to develop an instruction system for billiards for beginners using a projector-camera system. The direction and strength of shot are quite important in or- der to make the shot successful. The player's shooting stance is also important to shoot the cue-ball exactly. The direction and strength of shot and the proper shooting stanceare non-symbolic information which is difficult to send to the beginner. It is generally useful that the beginner easily ets these kinds of information. We use a projector to re- solve this problem. In this paper, we describe the method recognizing objects on the table, the method calculating a shooting path and shot difficulty, and the method showing the supporting information. We have confirmed experimental effectiveness of our support information.

DOI： 10.1145/1501750.1501752

Language：English   Publishing type：Research paper (other academic)  

We propose an approach to recognize time-series gesture patterns with Hierarchical Self-Organizing Map(HSOM). One of the key issue of the time-series pattern recognition is to absorb the time variant appropriately and to make cluters which include the same gesture class. In our approach. we arrange the SOM hierarchically. In each layer ofthe SOM time series patterns divided into some periods; postures, gesture elements and gestures. They are learned in each layer of HSOM. For example, postures are learned in the first layer, gesture elements are learned in the second layer and so on. Using the sparse code in the bottom layer, the SOM can perform time invarient recognition of the gesture elements and gestures.

DOI： 10.1109/icpr.2008.4761795

Language：English   Publishing type：Research paper (other academic)  

Hypercolumn model (HCM) is a neural network model previously proposed to solve image recognition problem. In this paper, we propose an improved version of HCM network and demonstrate its ability to solve face recognition problem. HCM network is a hierarchical model based on self-organizing map (SOM) that closely follows the organization of visual cortex and builds an increasingly complex and invariant feature representation. This invariance achieved by alternating between feature extraction and feature integration operation. To improve the recognition rate of HCM, we propose a variable dimension for each map in the feature extraction layer. The number of neurons in each map-side is decided automatically from training data. We demonstrate the performance of the approach using ORL face database.

DOI： 10.1109/IJCNN.2008.4633896

Language：English   Publishing type：Research paper (other academic)  

We propose an approach to improve the accuracy of estimating feature points of human body on a vision-based motion capture system (MCS) by using the Variable-density Self-Organizing Map (VDSOM). The VDSOM is a kind of SelfOrganizing Map (SOM) and has an ability to learn training samples incrementally. We let VDSOM learn 3-D feature points of human body when the MCS succeeded in estimating them correctly. On the other hand, one or more 3-D feature point could not be estimated correctly, we use the VDSOM for the other purpose. The SOM including VDSOM has an ability to recall a part of weight vector which have learned in the learning process. We use this ability to recall correct patterns and complement such incorrect feature points by replacing such incorrect feature points with them.

DOI： 10.1109/IJCNN.2008.4633912

Language：English   Publishing type：Research paper (scientific journal)  

We propose the use of a fast algorithm for estimating background models. This algorithm makes use of Parzen density estimation in non-stationary scenes. Each pixel has a probability density function this is used to approximate the value of pixels observed in a video sequence. Estimating this function quickly and accurately is important. In our approach, the probability density function is partially updated within the range of a window function based on the value observed. The model quickly adapts to changes in the scene and foreground objects can be robustly detected. Several experiments show the effectiveness of our approach.

DOI： 10.3169/itej.62.2045

Language：English   Publishing type：Research paper (other academic)  

Non-parametric representation of pixel intensity distribution is quite effective to construct proper background model and to detect foreground objects accurately. However, from the viewpoint of practical application, the computation cost of the distribution estimation should be reduced. In this paper, we present fast estimation of the probability density function (PDF) of pixel value using Parzen density estimation and foreground object detection based on the estimated PDF. Here, the PDF is computed by partially updating the PDF estimated at the previous frame, and it greatly reduces the computation cost of the PDF estimation. Thus, the background model adapts quickly to changes in the scene and, therefore, foreground objects can be robustly detected. Several experiments show the effectiveness of our approach.

DOI： 10.1109/AVSS.2007.4425366

Language：English   Publishing type：Research paper (other academic)  

We propose a fast algorithm to estimate background models using Parzen density estimation in non-stationary scenes. Each pixel has a probability density which approximates pixel values observed in a video sequence. It is important to estimate a probability density function fast and accurately. In our approach, the probability density function is partially updated within the range of the window function based on the observed pixel value. The model adapts quickly to changes in the scene and foreground objects can be robustly detected. In addition, applying our approach to cast-shadow modeling, we can detect moving cast shadows. Several experiments show the effectiveness of our approach.

Language：English   Publishing type：Research paper (scientific journal)  

We propose a new incremental learning method of Self-Organizing Map. There are three problems in the incremental learning of Self-Organizing Map; 1. neuron depletion, 2. forgetting previous training data, 3. keeping topology. Weights fixed neurons and weights semi-fixed neurons are very effective for the second problem. However the other problems remain. Therefore, we improve the incremental learning method with weights fixed neurons and weights semi-fixed neurons. Our approach can increment neurons effectively in the incremental learning process.

Language：English   Publishing type：Research paper (scientific journal)  

We propose an associatively learnable hypercolumn model (AHCM). A hyper-column model is a self-organized, competitive, and hierarchical multilayer neural network. It is derived from the neocognitron by replacing each S cell and C cell with a two-layer hierarchical self-organizing map. The HCM can recognize images with variant object size, position, orientation and spatial resolution. However, feature maps may integrate some features extracted in the lower layer even if the features are extracted from input data which belong to different categories. The learning algorithm of the HCM causes this problem because it is an unsupervised learning used by a self-organizing map. An associative learning method is therefore introduced, which is derived from the HCM by appending associative signals and associative weights to traditional input data and connection weights, respectively. The AHCM was applied to hand-shape recognition. We found that the AHCM could generate an appropriate feature map and higher recognition accuracy compared with the HCM.

DOI： 10.1007/s10015-006-0404-x

Language：English   Publishing type：Research paper (other academic)  

We propose a new incremental learning method of Self-Organizing Map. Basically, there are three problems in the incremental learning of Self-Organizing Map: 1. depletion of neurons, 2. oblivion of training data previously given, 3. destruction of topological relationship among training samples. Weight-fixed neurons and weight-quasi-fixed neurons are very effective for the second problem. However the other problems still remain. Therefore, we improve the incremental learning method with weight-fixed neurons and weight-quasi-fixed neurons. We solve the problems by introducing a mechanism to increase the number of neurons effectively in the incremental learning process.

Language：English   Publishing type：Research paper (other academic)  

We propose a method for create a background model in non-stationary scenes. Each pixel has a dynamic Gaussian mixture model Our approach can automatically change the number of Gaussians in each pixel. The number of Gaussians increases when pixel values often change because of Illumination change, object moving and so on. On the other hand, when pixel values are constant in a while, some Gaussians are eliminated or integrated. This process helps reduce computational time. We conducted experiments to investigate the effectiveness of our approach.

DOI： 10.1109/AVSS.2006.44

Language：English   Publishing type：Research paper (other academic)  

We are developing an Automatic Camera control system for Education: ACE, which captures a lecture using both a blackboard and a screen. ACE focuses on an oblect explained by a teacher. When this recording strategy is realized, it is necessary for ACE to extract a teacher's behavior and his/her explaining object. In this paper, we describe our algorithm to estimate a teacher's behavior by image processing and the camera control strategy to take suitable shots. We have applied ACE to recording a real lecture to validate it.

Event date： 2018.1

Language：Japanese  

Country：Japan  

Event date： 2018.1

Language：Japanese  

Country：Japan  

Event date： 2017.12

Language：Japanese  

Country：Japan  

Event date： 2017.11

Language：Japanese  

Country：Japan  

Event date： 2017.9

Language：Japanese  

Country：Japan  

Event date： 2017.8

Language：Japanese  

Country：Japan  

Event date： 2017.8

Language：Japanese  

Country：Japan  

Event date： 2017.8

Language：Japanese  

Country：Japan  

Event date： 2017.5

Language：Japanese  

Country：Japan  

Event date： 2016.4

Language：English   Presentation type：Oral presentation (general)  

Country：United Kingdom  

Event date： 2016.4

Language：English   Presentation type：Oral presentation (general)  

Country：United Kingdom  

Event date： 2016.4

Language：English   Presentation type：Oral presentation (general)  

Country：United Kingdom  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Presentation type：Oral presentation (general)  

Country：Japan  

Presentation type：Oral presentation (general)  

Country：Japan  

Country：France  

Country：France  

Country：France  

Presentation type：Oral presentation (general)  

Country：Japan  

Presentation type：Oral presentation (general)  

Country：Japan  

Presentation type：Oral presentation (general)  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Portugal  

Country：China  

Country：Japan  

Country：Japan  

Country：United States  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Korea, Republic of  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Korea, Republic of  

Country：Korea, Republic of  

Country：Korea, Republic of  

Country：Korea, Republic of  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Korea, Republic of  

Country：Korea, Republic of  

Country：Japan  

Country：United States  

Country：Japan  

Country：United States  

Country：United States  

Country：Japan  

Country：United States  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Italy  

Country：Italy  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：Japan  

Country：France  

Country：China  

Country：Japan  

Country：Japan  

Country：Japan  

Country：United Kingdom  

Country：China  

Country：Portugal  

Language：Japanese  

Country：Japan  

Country：Japan  

Country：Singapore  

Country：Japan  

Country：Korea, Republic of  

Country：Korea, Republic of  

Country：Japan  

Country：Malaysia  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：English  

Country：Italy  

Language：English  

Country：Taiwan, Province of China  

Language：English  

Country：Taiwan, Province of China  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：English  

Country：Germany  

Language：English  

Country：Germany  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：English  

Country：Canada  

Language：English  

Country：Australia  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：Japan  

Language：English  

Country：Japan  

Language：English  

Country：Chile  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：China  

Language：English  

Country：Italy  

Language：English  

Country：Japan  

Language：English  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：English  

Country：Austria  

Language：English  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：United States  

Language：English  

Country：Denmark  

Language：English   Presentation type：Oral presentation (general)  

Country：United States  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：Japanese  

Country：Japan  

Language：English  

Country：United Arab Emirates  

Language：Japanese  

Country：Japan  

Language：English  

Country：France  

Language：English   Presentation type：Oral presentation (general)  

Country：India  

Language：English  

Country：Japan  

Language：English  

Country：France  

Language：English  

Country：Japan  

Language：English  

Country：Japan  

Language：English  

Country：Mexico  

Language：Japanese  

Country：Japan  

Language：English  

Country：Iceland  

Language：Japanese  

Country：Japan  

Language：English  

Country：Korea, Republic of  

Language：English  

Country：Portugal  

Language：English  

Country：Portugal  

Language：English   Presentation type：Oral presentation (general)  

Country：Canada  

Language：English   Presentation type：Oral presentation (general)  

Country：Canada  

Language：English   Presentation type：Oral presentation (general)  

Country：Canada  

Language：English   Presentation type：Oral presentation (general)  

Country：Canada  

Language：English   Presentation type：Oral presentation (general)  

Country：Canada  

Language：English   Presentation type：Oral presentation (general)  

Country：Greece  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：English   Presentation type：Oral presentation (general)  

Country：Japan  

Language：Japanese  

Language：Japanese  

Language：Japanese  

Language：English  

Language：English  

Language：English  

Language：English  

Language：Japanese  

Language：Japanese  

Language：Japanese  

Language：English  

Language：English  

Language：English  

Language：English  

Language：Japanese  

Language：English  

Language：English  

Language：English  

Language：English  

Language：English  

Language：Japanese