Katsuya Ogata

I am a student at Osaka University majoring in information science, advised by Prof. Yuta Nakashima.

My research centers on computer vision and multimodal AI: video editing driven by subjective intent, visual concept reasoning, and the internal representations of multimodal models.

I also enjoy exchanging ideas with researchers abroad, e.g. through the ASPIRE project and ICLR 2026.

News

Nov. 2026Presenting our work on the Gavagai problem at IPSJ CVIM.
Oct. 2026Unveil the Mask is on arXiv.
Sep. 2026Virtual Encoders in Multimodal Transformers is on arXiv.
Jul. 2026MEDit-Bench is on arXiv.

Publications

International Conferences & Workshops

Virtual Encoders in Multimodal Transformers

Katsuya Ogata, Yuta Nakashima

ACCVW 2026

Unveil the Mask: Jointly-Optimized Diffusion Masked Pretraining for Appearance-Agnostic Operator Verification

Katsuya Ogata, Ryoya Yoshiyama, Hodaka Kawachi, Yuta Nakashima

ACCVW 2026

Preprints / Under Review

MEDit-Bench: A Dataset for Evaluating Message-Driven Narrative Video Editing

Katsuya Ogata, Zongshang Pang, Mayu Otani, Yuta Nakashima

arXiv 2026

Domestic Conferences (in Japanese)

Tackling the Gavagai Problem: Toward Concept Learning from a Single Positive and a Few Negatives in Japanese

Katsuya Ogata, Kazuya Nishimura, Yuta Nakashima

CVIM 2026

A Video Editing Dataset for Evaluating Intent-Driven Narrative Generation in Japanese

Katsuya Ogata, Zongshang Pang, Mayu Otani, Yuta Nakashima

MIRU 2026 Oral (33.5%)

Awards

Student Encouragement Award, MIRU 2026

2026

Interactive Presentation Award, MIRU 2026

2026

Σ Global Leader Award, School of Engineering Science, Osaka University

FY2025

Education

Osaka University
School of Engineering Science, Information and Computer Sciences

Activities

ASPIRE Project, Participant, international research exchange

ICLR 2026, Attendee

2026