Faculty Profiles - YONEZAWA Takuro

Papers - YONEZAWA Takuro

Division display 61 - 80 of about 307 ／ All the affair displays >>

Composite Image Generation Using Labeled Segments for Pattern-Rich Dataset without Unannotated Target Reviewed Open Access

Kano, K; Mori, Y; Higashiura, K; Hossain, T; Katayama, S; Urano, K; Yonezawa, T; Kawaguchi, N

COMPANION OF THE 2024 ACM INTERNATIONAL JOINT CONFERENCE ON PERVASIVE AND UBIQUITOUS COMPUTING, UBICOMP COMPANION 2024 page： 507 - 512 2024

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Ubicomp Companion 2024 Companion of the 2024 ACM International Joint Conference on Pervasive and Ubiquitous Computing

Although object detection technology using cameras offers potential for various applications, it incurs dataset creation costs to train new models where general-purpose models are ineffective, such as in industrial settings. We have previously developed a semi-automated annotation framework that employs optical flow and representation learning techniques to reduce human effort significantly. However, it was likely to cause unintended annotation omissions and mistakes compared to manual annotation. In this study, we propose a composite image generation approach to create omission-free and pattern-rich datasets. The proposed method synthesizes natural-looking images without unannotated targets by placing labeled foreground segments at their original positions on targetless background frames collected with the same fixed-point cameras. Evaluation with video footage in a logistics warehouse confirmed that improved dataset reliability led to higher model performance.

DOI： 10.1145/3675094.3678447

Open Access

Web of Science

Scopus

Additive Compositionality in Urban Area Embeddings Based on Human Mobility Patterns Reviewed

Naoki Tamura, Haru Terashima, Kazuyuki Shoji, Shin Katayama, Kenta Urano, Takuro Yonezawa, Nobuo Kawaguchi

Proceedings of the 32nd ACM International Conference on Advances in Geographic Information Systems page： 577 - 580 2024

OpenUAS: Embeddings of Cities in Japan with Anchor Data for Cross-city Analysis of Area Usage Patterns. Reviewed

Naoki Tamura, Kazuyuki Shoji, Shin Katayama, Kenta Urano, Takuro Yonezawa, Nobuo Kawaguchi

CoRR Vol. abs/2407.19872 2024

A Semi-Automated Ceiling Plan Generation System Using a Smartphone / an Omnidirectional Camera for Efficiency-Enhancing Reviewed Open Access

Techinical Papers of Annual Meeting the Society of Heating,Air-conditioning and Sanitary Engineers of Japan Vol. 2024.10 ( 0 ) page： 485 - 488 2024

Performance Evaluation of KNIME Low Code Platform in Deep Learning Study and Optimal Hyperparameter Tuning Reviewed

Thongkhome P., Yonezawa T., Kawaguchi N.

IEEE Region 10 Annual International Conference Proceedings TENCON page： 1373 - 1376 2024

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：IEEE Region 10 Annual International Conference Proceedings TENCON

A low-code platform is a software development environment that allows for the creation of applications through graphical user interfaces and configuration instead of traditional hand-coded computer programming. In this study, the application to classify a dataset of traffic sign images using the KNIME low-code deep learning development platform will be discussed to represents this software performance especial in term of model optimization processes. By creates the workflow to perform image preprocessing, create the CNN layer under KERAS sequential API and finding the best set of key hyperparameters among traditional KNIME build-in optimization algorithm including Brute force, Hill climbing, random search, Bayesian Optimization and black-box optimizer Optuna optimization algorithm under 3 types CNN architecture as simple CNN, Resnet-50 and VGG16 to classify traffic sign images. The result demonstrates that both grid search and random search optimization can be effective, while both Optuna and Bayesian optimization stands out as a powerful method due to its ability to efficiently explore the hyperparameter space and achieve superior results to meet 99% accuracy under simple CNN environment, but Optuna is significantly improve optimization times than Bayesian about 7 - 8 times. The KNIME low-code platform provides a user-friendly environment for developing and fine-tuning models to contribute the ongoing progress in machine learning and deep learning research development.

DOI： 10.1109/TENCON61640.2024.10903074

Scopus

Physiologically Expressive Robotic Hand as a Lifelike Presence Reviewed

Yonezawa, T; Meng, XS; Wan, X

SOCIAL COMPUTING AND SOCIAL MEDIA, PT III, SCSM 2024 Vol. 14705 page： 61 - 77 2024

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Lecture Notes in Computer Science Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics

In this study, we propose a robot hand that demonstrates a combination of various voluntary grips and involuntary physiological expressions on the skin as a tactile expression representing emotions. The robot hand is implemented using skin expressions such as variable gripping strength and speed, goosebumps, perspiration, and temperature, which can be tactually perceived by the user. The focus is on how human impressions of the robot change in human–robot touch communication, and we evaluated the combination of various conditions. By combining involuntary physiological expressions on the skin (goosebumps, sweating, and body temperature) with the grip of a hand robot as a tactile voluntary expression, we investigated the influence of the difference in expression between the robot’s fear–stress context and the normal context on the impression of the robot. We particularly focused on how the robot’s tangible expressions affect the user’s emotions, such as fears corresponding to the experimental context, and we analyzed whether this would lead to a sense of empathy, discomfort, etc. We confirmed that the user’s empathy, which could not be elicited by involuntary expressions alone, tended to be elicited by a combination of voluntary and involuntary expressions and that this tendency differed depending on the type of involuntary expressions.

DOI： 10.1007/978-3-031-61312-8_5

Web of Science

Scopus

Snuggling Robot at Different Positions Based on User's Facial Expressions Reviewed

Yonezawa, T; Matsuyama, H; Wan, X; Matsumura, N

2024 JOINT 13TH INTERNATIONAL CONFERENCE ON SOFT COMPUTING AND INTELLIGENT SYSTEMS AND 25TH INTERNATIONAL SYMPOSIUM ON ADVANCED INTELLIGENT SYSTEMS, SCIS&ISIS 2024

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：2024 Joint 13th International Conference on Soft Computing and Intelligent Systems and 25th International Symposium on Advanced Intelligent Systems Scis and Isis 2024

This study proposes a pet robot designed to enhance emotional interaction with users by adapting its behavior based on the recognition of facial expressions. The proposed system enables the robot to interpret a user's smile as an invitation to approach directly, while a sad expression prompts it to move closer in a comforting manner. Interaction results indicated that the robot's movement away from the user elicited feelings of loneliness and discomfort, whereas movements toward the user, alongside gentle snuggling and rotational gestures, fostered a sense of familiarity and reassurance.

DOI： 10.1109/SCISISIS61014.2024.10760154

Web of Science

Scopus

Trusted Inter-Reality基盤における利用主体に関するトラスト Invited Reviewed Open Access

金岡晃, 米澤拓郎

情報処理学会論文誌 Vol. 64 ( 12 ) page： 1590 - 1598 2023.10

CrowdFlowTransformer: Capturing Spatio-Temporal Dependence for Forecasting Human Mobility Invited

Choya T, Tamura N, Katayama S, Urano K, Yonezawa T, Kawaguchi N

PerCom Workshops page： 496 - 501 2023.3

Understanding Regional Characteristics Through EC Data Analysis. Reviewed

Kohei Yamaguchi, Kazuyuki Shoji, Naoki Tamura, Namiki Sakakura, Yuki Matsukura, Yu Hiranaga, Kohei Shimomura, Kenta Urano, Takuro Yonezawa, Nobuo Kawaguchi

HCI (26) Vol. 14036 page： 390 - 404 2023

Panoramic Ceiling Image Synthesis Method Prioritizing Fixture Outlines using an Omnidirectional Camera. Reviewed

Daiki Kohama, Yoshiteru Nagata, Kazushige Yasutake, Kenta Urano, Shin Katayama, Takuro Yonezawa, Nobuo Kawaguchi

Fourteenth International Conference on Mobile Computing and Ubiquitous Network(ICMU) page： 1 - 8 2023

Panoramic Ceiling Image Synthesis Method Prioritizing Fixture Outlines using an Omnidirectional Camera Reviewed

Kohama, D; Nagata, Y; Yasutake, K; Urano, K; Katayama, S; Yonezawa, T; Kawaguchi, N

2023 FOURTEENTH INTERNATIONAL CONFERENCE ON MOBILE COMPUTING AND UBIQUITOUS NETWORK, ICMU 2023

Open-Domain Dialogue Management Framework Across Multiple Device for Long-Term Interaction. Reviewed

Shin Katayama, Nozomi Hayashida, Kenta Urano, Takuro Yonezawa, Nobuo Kawaguchi

HCI (3) Vol. 14013 page： 354 - 365 2023

Digitization and Analysis Framework for Warehouse Truck Berth Reviewed

Yokoyama, K; Katayama, S; Urano, K; Yonezawa, T; Kawaguchi, N

2023 FOURTEENTH INTERNATIONAL CONFERENCE ON MOBILE COMPUTING AND UBIQUITOUS NETWORK, ICMU 2023

Digitization and Analysis Framework for Warehouse Truck Berth Reviewed

Yokoyama K., Katayama S., Urano K., Yonezawa T., Kawaguchi N.

2023 14th International Conference on Mobile Computing and Ubiquitous Network Icmu 2023 2023

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：2023 14th International Conference on Mobile Computing and Ubiquitous Network Icmu 2023

In distribution warehouses, various forms of data transformation are underway, such as the conversion of data related to receiving and inventory management. Distribution warehouses have designated areas known as truck berths, where trucks are parked for loading/unloading. While data collection is expected to enhance work efficiency at these truck berths, there is no data regarding the status of work in progress in the unloading space, and no system is in place to analyze the work. Consequently, it is unclear which tasks are problematic and how to enhance efficiency. To address this issue, this paper proposes a framework for analyzing the loading/unloading situation at truck berths with the objective of enhancing berth utilization efficiency and work efficiency. Specifically, we employed a fixed camera for object detection through instance segmentation. This approach enables us to understand the situation of trucks, unloading spaces, and workers. The training data were annotated using a large amount of real-world data, allowing for the measurement and analysis of the efficiency of loading/unloading operations. By calculating the amount of cargo based on the floor space, we assessed the efficiency of unloading operations per truck. Through the analysis of both efficient and inefficient operations, insights for enhancing work efficiency were provided.

DOI： 10.23919/ICMU58504.2023.10412228

Scopus

Gait-Robust Heading Estimation Using Horizontal Acceleration for Smartphone-based PDR Reviewed

Kano K., Yoshida T., Katayama S., Urano K., Yonezawa T., Kawaguchi N.

Ceur Workshop Proceedings Vol. 3581 2023

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Ceur Workshop Proceedings

This study tackles heading estimation for Pedestrian Dead Reckoning (PDR) with smartphones. In dealing with changes in the holding posture of smartphones, it works to consider the relationship between sensor orientation and heading. However, the existing methods lack robustness to various gaits, such as sideways and backward walking. Therefore, we propose a novel method considering various spatiotemporal features of horizontal acceleration with deep learning. The proposed method calculates horizontal acceleration in the global coordinate system from measured acceleration, gravitational acceleration, and rotation vector. Then, it inputs the horizontal acceleration over a certain period into a deep neural network model and predicts the unit vector directed to the mean heading during that period. We created a dataset covering multiple gaits and evaluated the method using four models: Convolutional Neural Network (CNN), Bidirectional Long Short-Term Memory (BiLSTM), DualCNN-LSTM, and DualCNN-Transformer. Consequently, we found that the proposed method was more robust to gaits than the existing methods, with the DualCNN-LSTM and DualCNN-Transformer models achieving the highest accuracy.

Scopus

Gait-Robust Heading Estimation Using Horizontal Acceleration for Smartphone-based PDR. Reviewed

Kazuma Kano, Takuto Yoshida, Shin Katayama, Kenta Urano, Takuro Yonezawa, Nobuo Kawaguchi

IPIN-WiP 2023

Mobility Link XR: Interspace Interaction System in Electric Wheelchair. Reviewed

Nozomi Hayashida, Hironori Shimosato, Kenta Urano, Takuro Yonezawa, Nobuo Kawaguchi

HCI (44) Vol. 1833 CCIS page： 314 - 320 2023

ER-Chat: A Text-to-Text Open-Domain Dialogue Framework for Emotion Regulation Reviewed

Shin Katayama, Shunsuke Aoki, Takuro Yonezawa, Tadashi Okoshi, Jin Nakazawa, Nobuo Kawaguchi

IEEE Transactions on Affective Computing Vol. 13 ( 4 ) page： 2229 - 2237 2022.10

Dialogue Response Generation for Text-Based Dialogue Systems with Emotion Regulation

KATAYAMA Shin, YONEZAWA Takuro, KAWAGUCHI Nobuo

Journal of Japan Society for Fuzzy Theory and Intelligent Informatics Vol. 34 ( 3 ) page： 568 - 578 2022.8

Language：English Publishing type：Research paper (scientific journal) Publisher：Japan Society for Fuzzy Theory and Intelligent Informatics

<p>Dialogue systems that handle emotions have been shown to improve user satisfaction and increase positive interactions, and are expected to play a role as a digital partner that accompanies humans. In order for dialogue systems to recognize user emotions and express their own emotions, emotion regulation methods that select appropriate emotions from the context of the dialogue are indispensable. In this paper, we propose a text-based dialogue framework that performs emotion regulation, which estimates appropriate emotions during response. Our proposed method estimates appropriate emotions by considering the semantic and emotional context of the user’s text, and generates dialogue responses with emotional expressions using a neural network. The effectiveness of the proposed method was demonstrated through automatic evaluation using an evaluation metrics for the quality of dialogue generation and human evaluation by 100 subjects collected through a crowdsourcing.</p>

DOI： 10.3156/jsoft.34.3_568

CiNii Research

　 PREV - NEXT 　

<YONEZAWA Takuro>

Papers - YONEZAWA Takuro