Activity
Laboratory news covering research presentations, publications, awards, invited talks, academic service, outreach, collaboration, and internal activities.
Recent Topics
2026.08.06PresentationDomestic ConferenceComputer Vision
Scheduled to present 2D human pose estimation research at MIRU 2026
Research by Wataru Hashizume and Hiroyasu Usami on 2D human pose estimation via compressed token learning with structural consistency is listed as an interactive presentation at MIRU 2026. The presentation is scheduled for Interactive Session IS3 on August 6, 13:00–15:00.
- 2D Human Pose Estimation via Compressed Token Learning with Structural Consistency
- Authors: Wataru Hashizume, Hiroyasu Usami
- Presentation ID: IS3-171, Paper ID: 366
Related Links
2026.06.14PublicationPresentationGenerative AICollaboration
An arXiv preprint on LLM judge reliability evaluation is now available
The preprint “LLM Judges Have Dark Current: A Psychometric Datasheet for LLM-as-a-Judge Evaluation” is now available on arXiv. The work treats LLM-as-a-judge systems as measurement instruments rather than scalar accuracy or agreement devices, and proposes a Judge Datasheet protocol for measuring dark current, positional bias, target sensitivity, and criterion shift.
- arXiv:2606.15610 [cs.CL]
- Authors: Hiroyasu Usami, Keisuke Hara, Ayato Tsuboi, Naohiko Matsuda
- Subject classes: cs.CL, astro-ph.IM, cs.AI, cs.LG
Related Links
2026.06.04PublicationMedical Imaging
A paper on an AI arteriosclerosis diagnosis system was published in Journal of Comprehensive Engineering
A paper on an AI arteriosclerosis diagnosis system using RLHF and adversarial attack was published in Vol. 38 of Journal of Comprehensive Engineering. The study evaluates the robustness of IMT extraction from carotid ultrasound images under adversarial perturbations, focusing on whether attention remains on physician-relevant structures.
- AI Arteriosclerosis Diagnosis System Using RLHF and Adversarial Attack: Quantitative Evaluation of IMT Extraction Robustness by Adversarial Learning
- Author: Hiroyasu Usami
- Journal of Comprehensive Engineering, Vol. 38, pp. 1-11, March 2026
Related Links
2026.04.01Academic ServiceIPSJ
Hiroyasu Usami became a representative member of the Information Processing Society of Japan
Hiroyasu Usami became a representative member of the Information Processing Society of Japan for fiscal year 2026. He also serves the IPSJ Tokai Branch as an operating committee member and secretary, and is an IPSJ senior member.
Related Links
2026.03.07OutreachEducation
Served as a poster judge for the 8th Junior and Senior High School Informatics Research Contest
Hiroyasu Usami served as a judge for informatics research poster presentations by junior and senior high school students, supporting exploratory research and education in informatics.
Related Links
2026.03.02International ConferenceMedical Imaging
Presented medical image segmentation research at BIOIMAGING 2026
We presented research on Stable Diffusion-based data augmentation for medical image segmentation. The study addresses generative-model use and evaluation design for small medical image datasets.
- Anatomy-Preserving Diffusion-Based Data Augmentation for Medical Image Segmentation
- Authors: Hiroki Tominari, Natsuki Nakayama, Naoko Arakawa, Hiroyasu Usami
Related Links
2026.02.13AwardEducation
Received an Excellence Award at the NIBB-Chubu University Hackathon 2025
Ayato Tsuboi, Keisuke Hara, Atsuki Kawaguchi, and Yuri Yokochi received an Excellence Award at the NIBB Hackathon 2025. Their project connected AI with biological images by reconstructing 3D models from citizen-science photographs.
- Result announcement: February 13, 2026
- Project title: Kawasemi is All You Need, second round: rapid 3D kingfisher morphology reconstruction using InstantMesh
Related Links
2026.01.23PresentationIoT and GNN
Presented building IoT simulation research at the EST Technical Committee
We presented research on computational optimization in building IoT simulation and physics-informed STGNNs, examining applications of spatiotemporal graph neural networks.
- O(V) Computational Optimization and Physics-informed STGNN for Building IoT Simulation
- Authors: Koichiro Tajima, Hiroyasu Usami
Related Links
2026.01.12OutreachEducation
Helped organize the Interesting Science Class at Nagoya University
As part of IPSJ Tokai Branch activities, Hiroyasu Usami helped organize the Interesting Science Class at Nagoya University, an outreach event for science and informatics education.
Related Links
2025.12.07PresentationDomestic Conference
Presented three studies at WiNF 2025
We presented three studies on outdoor image processing, educational support, and evaluation of image generative AI, showing the laboratory breadth from image understanding to generative AI applications.
- Rain-effect Removal and Weather Estimation for Outdoor Images Using Rain Artifact Separation Learning
- Authors: Manato Kikuta, Wataru Hashizume, Hiroyasu Usami
- A Stabilization Method for Automatic Educational Question Generation Based on Self-consistency in Chain-of-Thought Reasoning
- Authors: Nayuta Shimotsuji, Hiroyasu Usami, Jin Nakane
- Verification of Semantic Biosynthesis in Image Generative AI Through a Modality Conversion-based Generation and Verification Cycle
- Authors: Keima Izuhara, Hiroki Tominari, Hiroyasu Usami
Related Links
2025.11.05Intellectual PropertyMedical Imaging
Obtained a patent related to endoscopic image conversion
We obtained a patent for an endoscopic image conversion device, system, and program. This result supports social implementation of medical image processing technology.
- Patent No. 7769330
Related Links
2025.09.18PresentationDomestic Conference
Presented two studies at the Tokai-Section Joint Conference
We presented studies on communication-log analysis and retinal image segmentation. The presentations covered time-series analysis using GNNs and data augmentation for medical image segmentation.
- Application of a Fully Connected Spatio-Temporal GNN Structure for Day-of-week Classification Based on Communication Logs
- Authors: Koichiro Tajima, Yuki Nakashima, Hiroyasu Usami
- Data Augmentation for Retinal Image Segmentation Based on Ground-truth Mask Images
- Authors: Hiroki Tominari, Hiroyasu Usami
Related Links
2025.09.18Conference OrganizationDomestic Conference
Organized and chaired sessions at the Tokai-Section Joint Conference
At the 2025 Tokai-Section Joint Conference on Electrical, Electronics, Information, and Related Engineering, Hiroyasu Usami organized and chaired the OS6 planned session on natural speech dialogue systems and chaired a Media Processing session.
- OS6: Development and Challenges of Natural Speech Dialogue Systems, organizer and chair
- September 18, 13:30-15:15, Room H
- Media Processing session, chair
- September 19, 14:45-16:45, Room H (C3-301)
Related Links
2025.07.30PresentationComputer Vision
Presented inverse rendering research at MIRU 2025
We presented research on inverse rendering using single-view polarization cues and shape guidance under ambient light, addressing estimation of object shape and reflectance.
- Inverse Rendering Using Single-view Polarization Cues and Shape Guidance Under Ambient Light
- Authors: Ayato Tsuboi, Hiroyasu Usami
Related Links
2025.03.15OutreachEducation
Served as a poster judge for the 7th Junior and Senior High School Informatics Research Contest
Hiroyasu Usami served as a judge for informatics research poster presentations by junior and senior high school students.
Related Links
2025.02.20International ConferenceMedical Imaging
Presented carotid ultrasound image analysis research at BIOIMAGING 2025
We presented a lightweight IMT region extraction method for mobile carotid ultrasound imaging using MobileNetV3 and SegNet.
- Efficient Automatic IMT Region Extraction Using MobileNetV3 and SegNet for Mobile Carotid Ultrasound Imaging
- Authors: Hiroyasu Usami, Natsuki Nakayama, Naoko Arakawa
Related Links
2025.01.20AwardMedical Imaging
Hiroki Tominari received the 2025 Tokai-Section Joint Conference Encouragement Award
Hiroki Tominari received an Encouragement Award for his presentation on data augmentation for retinal image segmentation at the 2025 Tokai-Section Joint Conference.
- Title: Data augmentation for retinal image segmentation based on ground-truth masks
- Awardee: Hiroki Tominari
Related Links
2025.01.13OutreachEducation
Helped organize the 2025 Interesting Science Class at Nagoya University
As part of IPSJ Tokai Branch activities, Hiroyasu Usami helped organize the Interesting Science Class at Nagoya University for local pupils and students.
2024.09.24AwardEducation
Won a Jury Award and a Corporate Award at HACK-Aichi 2024
Yuui Ito, Daito Suzuki, Hiroki Tominari, Ayato Tsuboi, and Jin Nakane won a Jury Award and a Corporate Award at HACK-Aichi, an inter-university hackathon hosted by Aichi Prefecture.
- Project title: Biz Mate
Related Links
2024.08.28Conference OrganizationIPSJ
Organized a special session at the Tokai-Section Joint Conference
Hiroyasu Usami organized a special session on understanding and applying AI technologies and chaired a Learning Support session at the 2024 Tokai-Section Joint Conference on Electrical, Electronics, Information, and Related Engineering.
- Special session: organizer for a session on understanding and applying AI technologies
- August 29, H venue, Engineering 106, 10:00-12:00, Learning Support: session chair
Related Links
2024.08.05PresentationMedical Imaging
Presented IMT segmentation research at the 56th Tokai Fuzzy Workshop
We presented research on IMT region segmentation in carotid ultrasound images using SegNet, addressing lightweight segmentation for medical image analysis.
- SegNet-based IMT Region Segmentation in Carotid Ultrasound Images
- Authors: Mizuki Doi, Hiroyasu Usami, Natsuki Nakayama, Naoko Arakawa
Related Links
2024.07.25Regional CollaborationEducation
Gave a data-utilization seminar at the Kasugai Chamber of Commerce and Industry
Hiroyasu Usami lectured on data utilization and AI for local businesses at a seminar hosted by the commercial division of the Kasugai Chamber of Commerce and Industry.
2024.04.01Academic ServiceIPSJ
Participating in IPSJ Tokai Branch operations
Hiroyasu Usami serves as an operating committee member and secretary of the IPSJ Tokai Branch. Since 2024, he has also been an IPSJ senior member.
Related Links
2024.02.18PresentationGenerative AIAward
Presented image generative AI research and received an Excellent Presentation Award at the 55th Tokai Fuzzy Workshop
We presented research on purpose-oriented prompts for image generative AI, and Reiya Fukuoka received an Excellent Presentation Award.
- A Study of Purpose-oriented Prompts for Image Generative AI
- Authors: Reiya Fukuoka, Hiroki Tominari, Hiroyasu Usami, Yuji Iwahori
- Award: Excellent Presentation Award, awardee: Reiya Fukuoka
Related Links
2023.12.10PresentationDomestic Conference
Presented two studies at WiNF 2023
We presented research on dangerous worker-action recognition and prompt evaluation for Stable Diffusion-based image generative AI.
- Image Processing-based Support System for Recognizing Dangerous Worker Actions and Preventing Machine Accidents
- Authors: Wataru Hashizume, Hiroyasu Usami, Yusuke Noguchi, Hiroyuki Tsuchida, Yuji Iwahori
- Research and Evaluation of Effective Prompts for Stable Diffusion-based AI Image Generation Tools
- Authors: Hiroki Tominari, Reiya Fukuoka, Hiroyasu Usami
Related Links
2023.11.30Invited TalkOutreach
Gave an invited talk in a special session at MWE 2023
Hiroyasu Usami gave an invited talk in the Microwave Workshop and Exhibition 2023 special session on future and foundational technologies, reviewing the technical transition from AI to machine learning and deep learning.
Related Links
2023.04.01Collaboration
Joined a collaborative mHealth research project in rural Thailand
Hiroyasu Usami joined a KAKENHI research project on developing an mHealth application to improve follow-up for patients with lifestyle-related diseases in rural Thailand.
2023.01.27PresentationComputer Vision
Presented biomass ash sample image-analysis research at the EST Technical Committee
We presented research on extracting sample regions from biomass ash dissolution image data using time-series information.
- Sample-region Extraction in Biomass Ash Dissolution Image Data Using Time-series Information
- Authors: Haruki Yamane, Hiroyasu Usami, Yuri Suzuki, Yoshihiko Ninomiya
2022.12.17PresentationDomestic Conference
Presented four studies at WiNF 2022
We presented four studies on generative AI, edge computing, medical image analysis, and coal image analysis.
- A Text-to-image Model Considering Latent Information Specific to Japanese
- Authors: Reiya Fukuoka, Daiki Yamazaki, Haruki Yamane, Hiroyasu Usami, Yuri Suzuki
- Adaptive Virtualization Environment Using Edge Computing
- Authors: Yuto Hayashi, Haruki Yamane, Hiroyasu Usami
- Segmentation of Laterally Spreading Tumors Using an Attention Mechanism
- Authors: Mizuki Doi, Hiroyasu Usami, Yuji Iwahori, Naotaka Ogasawara, Kunio Kasugai
- Coal Image Segmentation Using an Attention Mechanism Considering Relationships Between Coal and Included Minerals
- Authors: Koichiro Tajima, Mizuki Doi, Hiroyasu Usami, Yoshihiko Ninomiya
Related Links
2022.09.29Invited TalkOutreach
Gave an invited talk at an information exchange meeting hosted by Tohoku Electric Power Engineering and Construction and Tohoku Ryokka Environmental Preservation
Hiroyasu Usami gave a lecture on the technical transition from artificial intelligence to machine learning and deep learning from the perspective of image processing.
2022.04.01Lab Activity
Usami Laboratory was founded
Usami Laboratory was founded on April 1, 2022, starting research activities around light, image understanding, and medical applications.