Yoshifumi Nishida
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Yoshifumi Nishida has a total of 58 co-authors. Here are the top co-authors by number of shared publications:
Hiroshi Mizoguchi
Tokyo Univ. of Science
17 shared publications- Koji Kitamura
National Institute of Advanced Industrial Science and Technology, Tokyo, Japan
17 shared publications - Yoichi Motomura
AIST
9 shared publications Tatsuhiro Yamanaka
Ryokuen Children's Clinic and Safe Kids Japan, Japan
7 shared publications- Yoshinori Koizumi
Designing Everyday Life Function and Social System Team, Digital Human Research Center, National Institute of Advanced Industrial Science and Technology (AIST), 2-3-26 Aomi, Koto-ku, Tokyo 135-0064, Japan
7 shared publications - Tachio Takano
Designing Everyday Life Function and Social System Team, Digital Human Research Center, National Institute of Advanced Industrial Science and Technology (AIST), 2-3-26 Aomi, Koto-ku, Tokyo 135-0064, Japan
6 shared publications
Latest Publications
- Open Access
- 14 June 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 43 (1) , 269-283
RotationNet for Joint Object Categorization and Unsupervised Pose Estimation from Multi-View Images
We propose a Convolutional Neural Network (CNN)-based model "RotationNet," which takes multi-view images of an object as input and jointly estimates its pose and object category. Unlike previous approaches that use known viewpoint labels for training, our method treats the viewpoint labels as latent variables, which are learned in an unsupervised manner during the training using an unaligned object dataset. RotationNet uses only a partial set of multi-view images for inference, and this property makes it useful in practical scenarios where only partial views are available. Moreover, our pose alignment strategy enables one to obtain view-specific feature representations shared across classes, which is important to maintain high accuracy in both object categorization and pose estimation. Effectiveness of RotationNet is demonstrated by its superior performance to the state-of-the-art methods of 3D object classification on 10- and 40-class ModelNet datasets. We also show that RotationNet, even trained without known poses, achieves comparable performance to the state-of-the-art methods on an object pose estimation dataset. Furthermore, our object ranking method based on classification by RotationNet achieved the first prize in two tracks of the 3D Shape Retrieval Contest (SHREC) 2017. Finally, we demonstrate the performance of real-world applications of RotationNet trained with our newly created multi-view image dataset using a moving USB camera.
- 14 June 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 43 (1) , 269-283
Open AccessRotationNet for Joint Object Categorization and Unsupervised Pose Estimation from Multi-View Images
We propose a Convolutional Neural Network (CNN)-based model "RotationNet," which takes multi-view images of an object as input and jointly estimates its pose and object category. Unlike previous approaches that use known viewpoint labels for training, our method treats the viewpoint labels as latent variables, which are learned in an unsupervised manner during the training using an unaligned object dataset. RotationNet uses only a partial set of multi-view images for inference, and this property makes it useful in practical scenarios where only partial views are available. Moreover, our pose alignment strategy enables one to obtain view-specific feature representations shared across classes, which is important to maintain high accuracy in both object categorization and pose estimation. Effectiveness of RotationNet is demonstrated by its superior performance to the state-of-the-art methods of 3D object classification on 10- and 40-class ModelNet datasets. We also show that RotationNet, even trained without known poses, achieves comparable performance to the state-of-the-art methods on an object pose estimation dataset. Furthermore, our object ranking method based on classification by RotationNet achieved the first prize in two tracks of the 3D Shape Retrieval Contest (SHREC) 2017. Finally, we demonstrate the performance of real-world applications of RotationNet trained with our newly created multi-view image dataset using a moving USB camera.
- 1 June 2019
- Published by Elsevier in Safety Science
- Vol. 115, 199-207
A matrix analysis of carrying device-related injuries in preschool children
- 1 June 2019
- Published by Elsevier in Safety Science
- Vol. 115, 199-207
A matrix analysis of carrying device-related injuries in preschool children
- Open Access
- 16 April 2019
- Published by IntechOpen
Living Function-Resilient Society in the Centenarian Era: Living Safety Technology Based on Connective, Artificial Intelligence
In the centenarian era, it has become even more imperative to address the physical and cognitive changes faced by children, the elderly, and disabled persons. We need a “living function resilient society,” which ensures they enjoy safe living environments in ways that allow them to maintain active social participation levels despite the changes. To build such a society, more attention should be paid to problems: diversity in life function and intervention needs, gap between efficacy and effectiveness, fragmentation of living data and support service and variety in privacy exposure. To deal with these issues, this chapter describes a new approach referred to as “connective AI” that allows individual lives to be connected with each other and efficacy to be scaled to effectiveness by computerizing places of living in accordance with the private policy of individual facilities and connecting them with each other through a network. As a concrete example of connective AI, this chapter introduces smart living labs that are developed by the National Institute of Advanced Industrial Science and Technology (AIST) in cooperation with children’s hospitals, rehabilitation hospitals, intensive care homes for the elderly, and private homes.
- 16 April 2019
- Published by IntechOpen
Open AccessLiving Function-Resilient Society in the Centenarian Era: Living Safety Technology Based on Connective, Artificial Intelligence
In the centenarian era, it has become even more imperative to address the physical and cognitive changes faced by children, the elderly, and disabled persons. We need a “living function resilient society,” which ensures they enjoy safe living environments in ways that allow them to maintain active social participation levels despite the changes. To build such a society, more attention should be paid to problems: diversity in life function and intervention needs, gap between efficacy and effectiveness, fragmentation of living data and support service and variety in privacy exposure. To deal with these issues, this chapter describes a new approach referred to as “connective AI” that allows individual lives to be connected with each other and efficacy to be scaled to effectiveness by computerizing places of living in accordance with the private policy of individual facilities and connecting them with each other through a network. As a concrete example of connective AI, this chapter introduces smart living labs that are developed by the National Institute of Advanced Industrial Science and Technology (AIST) in cooperation with children’s hospitals, rehabilitation hospitals, intensive care homes for the elderly, and private homes.
- Open Access
RotationNet: Joint Object Categorization and Pose Estimation Using Multiviews from Unsupervised Viewpoints
We propose a Convolutional Neural Network (CNN)-based model "RotationNet," which takes multi-view images of an object as input and jointly estimates its pose and object category. Unlike previous approaches that use known viewpoint labels for training, our method treats the viewpoint labels as latent variables, which are learned in an unsupervised manner during the training using an unaligned object dataset. RotationNet is designed to use only a partial set of multi-view images for inference, and this property makes it useful in practical scenarios where only partial views are available. Moreover, our pose alignment strategy enables one to obtain view-specific feature representations shared across classes, which is important to maintain high accuracy in both object categorization and pose estimation. Effectiveness of RotationNet is demonstrated by its superior performance to the state-of-the-art methods of 3D object classification on 10- and 40-class ModelNet datasets. We also show that RotationNet, even trained without known poses, achieves the state-of-the-art performance on an object pose estimation dataset. The code is available on https://github.com/kanezaki/rotationnet
Open AccessRotationNet: Joint Object Categorization and Pose Estimation Using Multiviews from Unsupervised Viewpoints
We propose a Convolutional Neural Network (CNN)-based model "RotationNet," which takes multi-view images of an object as input and jointly estimates its pose and object category. Unlike previous approaches that use known viewpoint labels for training, our method treats the viewpoint labels as latent variables, which are learned in an unsupervised manner during the training using an unaligned object dataset. RotationNet is designed to use only a partial set of multi-view images for inference, and this property makes it useful in practical scenarios where only partial views are available. Moreover, our pose alignment strategy enables one to obtain view-specific feature representations shared across classes, which is important to maintain high accuracy in both object categorization and pose estimation. Effectiveness of RotationNet is demonstrated by its superior performance to the state-of-the-art methods of 3D object classification on 10- and 40-class ModelNet datasets. We also show that RotationNet, even trained without known poses, achieves the state-of-the-art performance on an object pose estimation dataset. The code is available on https://github.com/kanezaki/rotationnet
- 1 January 2018
- Published by INSTICC
- p. 361-368
“Change the Changeable” Framework for Implementation Research in Health
In recent years, a “participatory” system of community members and researchers has become a key factor to obtaining the best outcomes to create a better world. The authors previously suggested the problem structure change theory to find how to address social issues. In this paper, we suggest the “change the changeable framework” with three approaches for community-based program implementation: 1) Share the value of the change the changeable, 2) Systematize the process of a problem structure change, and 3) Build a sustainable system and discuss a case study based on the framework. In the case study, we conducted a safety education curriculum for injury prevention that consisted of what is injury prevention, playground safety, indoor safety, and photovoice project. We introduce an injury surveillance system and a risk recognition system as new tools for the collaboration of human intelligence and artificial intelligence.
- 1 January 2018
- Published by INSTICC
- p. 361-368
“Change the Changeable” Framework for Implementation Research in Health
In recent years, a “participatory” system of community members and researchers has become a key factor to obtaining the best outcomes to create a better world. The authors previously suggested the problem structure change theory to find how to address social issues. In this paper, we suggest the “change the changeable framework” with three approaches for community-based program implementation: 1) Share the value of the change the changeable, 2) Systematize the process of a problem structure change, and 3) Build a sustainable system and discuss a case study based on the framework. In the case study, we conducted a safety education curriculum for injury prevention that consisted of what is injury prevention, playground safety, indoor safety, and photovoice project. We introduce an injury surveillance system and a risk recognition system as new tools for the collaboration of human intelligence and artificial intelligence.
- 1 January 2017
- Published by Japan Society of Mechanical Engineers in The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec)
- Vol. 2017, 2P2-G11
A system for detecting fatal situation using RGB-D cameras for preventing children falls from balconies
Over 100 children die of falls from balconies for five years in Tokyo. This paper proposes a new system for detecting fatal situation using RGB-D camera for preventing children falls from balconies. The system has two functions. The first is to detect a child inside a target balcony by estimating his or her age from the length of the body trunk. The second is to detect the places that the child can climb by evaluating shape of objects in the balcony. We conducted experiments at the AIST’s living laboratory and an ordinary apartment, and confirmed the effectiveness of the functions.
- 1 January 2017
- Published by Japan Society of Mechanical Engineers in The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec)
- Vol. 2017, 2P2-G11
A system for detecting fatal situation using RGB-D cameras for preventing children falls from balconies
Over 100 children die of falls from balconies for five years in Tokyo. This paper proposes a new system for detecting fatal situation using RGB-D camera for preventing children falls from balconies. The system has two functions. The first is to detect a child inside a target balcony by estimating his or her age from the length of the body trunk. The second is to detect the places that the child can climb by evaluating shape of objects in the balcony. We conducted experiments at the AIST’s living laboratory and an ordinary apartment, and confirmed the effectiveness of the functions.
- 1 January 2016
- Published by Japan Society of Mechanical Engineers in The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec)
- Vol. 2016, 2A1-10a4
Design and evaluation method of support equipment based on living system understanding
- 1 January 2016
- Published by Japan Society of Mechanical Engineers in The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec)
- Vol. 2016, 2A1-10a4
Design and evaluation method of support equipment based on living system understanding
- 1 April 2015
- Published by Elsevier in Forensic Science International
- Vol. 249, 101-106
Statistical analysis of biomechanical properties of the adult sagittal suture using a bending method in a Japanese forensic sample
- 1 April 2015
- Published by Elsevier in Forensic Science International
- Vol. 249, 101-106
Statistical analysis of biomechanical properties of the adult sagittal suture using a bending method in a Japanese forensic sample
- Open Access
- 1 January 2015
- Published by Elsevier in Procedia Manufacturing
- Vol. 3, 4906-4913
Development of a Nursing Care Support System that Seamlessly Monitors Both Bedside and Indoor Locations
- 1 January 2015
- Published by Elsevier in Procedia Manufacturing
- Vol. 3, 4906-4913
Open AccessDevelopment of a Nursing Care Support System that Seamlessly Monitors Both Bedside and Indoor Locations