I am an Assistant Professor at The University of Tokyo, Japan. My research spans multimodal AI and data-centric AI. My recent work focuses on leveraging generative AI for education (e.g., LLM-code-detector), evaluation (LMM-as-a-judge), and applications in Japanese cultural contexts (JMMMU, MangaLMM). I previously worked on OCR, including multilingual text recognition and synthetic visual text generation (TRBA, COO, CLL-STR).

CV | email | Google Scholar | LinkedIn | Github

Work experience

Education

Publications

(*: Equal contribution)

MangaVQA and MangaLMM: A Benchmark and Specialized Model for Multimodal Manga Understanding

• Jeonghun Baek*, Kazuki Egashira*, Shota Onohara*, Atsuyuki Miyai*, Yuki Imajuku, Hikaru Ikuta, Kiyoharu Aizawa
• European Chapter of the Association for Computational Linguistics (EACL), Findings, 2026 [Project page] [Paper] [Code]
• Extended abstract version@International Conference on Computer Vision (ICCV) COMIQ Workshop, 2025 [Slides]

JMMMU-Pro: Image-based Japanese Multi-discipline Multimodal Understanding Benchmark via Vibe Benchmark Construction

• Atsuyuki Miyai, Shota Onohara, Jeonghun Baek, Kiyoharu Aizawa
• arXiv preprint 2025 [Project page]

MaskingAgent: Preventing LLM Tutor from Providing Full Solutions in Python Programming Courses

• Jeonghun Baek, Tetsuro Yamazaki, Akimasa Morihata, Junichiro Mori, Yoko Yamakata, Kenjiro Taura, Shigeru Chiba
• ACM Special Interest Group on Computer Science Education (SIGCSE) Technical Symposium, poster, 2026 [Paper] [Poster]

LLM-Based Explainable Detection of LLM-Generated Code in Python Programming Courses

• Jeonghun Baek, Tetsuro Yamazaki, Akimasa Morihata, Junichiro Mori, Yoko Yamakata, Kenjiro Taura, Shigeru Chiba
• ACM Special Interest Group on Computer Science Education (SIGCSE) Technical Symposium, paper (acceptance rate 30.3%, 174/574), 2026 [Paper] [Slide]

Exploring LMM-as-a-Judge for Image Harmonization Evaluation

• Jeonghun Baek*, Eunchung Noh*
• International Conference on Computer Vision (ICCV) UniLight Workshop, 2025

Enhancing Safety Judgment on LLM Responses via Text-to-Image Generation

• Eunchung Noh*, Jeonghun Baek*
• International Conference on Computer Vision (ICCV) WiCV Workshop, 2025

Toward Responsible Federated Large Language Models: Leveraging a Safety Filter and Constitutional AI

• Eunchung Noh*, Jeonghun Baek*
• arXiv preprint 2025 [Paper]

Harnessing PDF Data for Improving Japanese Large Multimodal Models

• Jeonghun Baek, Akiko Aizawa, Kiyoharu Aizawa
• Association for Computational Linguistics (ACL), Findings, 2025 [Paper] [Code (placeholder)]

JMMMU: A Japanese Massive Multi-discipline Multimodal Understanding Benchmark for Culture-aware Evaluation

• Shota Onohara*, Atsuyuki Miyai*, Yuki Imajuku*, Kazuki Egashira*, Jeonghun Baek*, Xiang Yue, Graham Neubig, Kiyoharu Aizawa
• Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL), 2025, and Neural Information Processing Systems (NeurIPS) EvalEval Workshop (oral), 2024 [Project page]