Hi, I’m Xuhan (盛 栩涵), an incoming Ph.D. student at the Institute of Science Tokyo (formerly Tokyo Institute of Technology), advised by Prof. Rio Yokota. I received my Master’s degree from Peking University, advised by Prof. Jian ZHANG, and my Bachelor’s degree in Artificial Intelligence from Dalian University of Technology, advised by Prof. Xu JIA.

My current research interests are in world models and MLLM.

I am always open to academic and industry collaborations. Feel free to reach out.

🔥 News

• 2026.10: 🎓 I will begin my Ph.D. studies at the Institute of Science Tokyo.
• 2026.03: 💼 Started my internship at Tencent IEG (Game AI Engine Department).
• 2025.09: 🎉 Collaboration: 1 paper accepted in PRCV 2025.
• 2025.08: 🎉 Collaboration: 1 paper accepted in IJCV 2025.
• 2025.06: 🎉 Collaboration: 1 paper accepted in ICME 2025 for oral presentation.
• 2025.05: 💼 Started my internship at OPPO Research Institute (AI Talent Program).
• 2024.09: 🎉 First-author: 1 paper (“OmniSSR”) accepted in ECCV 2024 for oral presentation.
• 2023.06: 🏆 Collaboration: 1 work won the “NTIRE 2023 Challenge on 360deg Omnidirectional Image Super-Resolution track” championship.

💼 Industry Experience

Tencent, IEG, Game AI Engine Department

2026.03 - Present · Intern

• Qwen-Image-based image enhancement for game content and production-level visual assets.
• Building data-pipeline operators for video generative models; developed a streaming shot-segmentation operator that splits long videos into single-shot clips.

OPPO Research Institute

2025.05 - 2025.09 · Imaging Algorithm Engineer Intern

Worked on vision-language models for assisting AI-based image enhancement. Advisor: Prof. Lei Zhang.

🔬 Research Interests

• World models for interactive long-video generation, with research focused on post-training and distillation for acceleration.
• Diffusion-based restoration and generation for panoramic visual content.

📑 Publications

(*: indicates equal contribution; #: indicates corresponding author)

Diffusion Models for Image & Video Generation and Restoration

Under Review

RealOSR: Latent Unfolding Boosts Diffusion-based Real-world Omnidirectional Image Super-Resolution

Xuhan Sheng*, Runyi Li*, Bin Chen, Weiqi Li, Xu Jiang, Jian Zhang#

Under Review

One-step diffusion for real-world 360° super-resolution via latent-space unfolding with domain-alignment and degradation-aware modules.

Paper

IJCV 2025

OmniDrag: Enabling Motion Control for Omnidirectional Image-to-Video Generation

Weiqi Li, Shijie Zhao, Chong Mou, Xuhan Sheng, et al.

IJCV 2025

First diffusion framework bringing precise drag-style motion control to omnidirectional image-to-video generation on the spherical domain.

Paper

ICME 2025 Oral

Label-guided Facial Retouching Reversion

Guanhua Zhao*, Yu Gu*, Xuhan Sheng, Yujie Hu, Jian Zhang#

ICME 2025 Oral Presentation

Label-guided diffusion that reverses skin smoothing, eye enlargement and face slimming, recovering authentic portraits with identity preserved.

Paper

ECCV 2024 Oral

OmniSSR: Zero-shot Omnidirectional Image Super-Resolution using Stable Diffusion Model

Runyi Li*, Xuhan Sheng*, Weiqi Li, Jian Zhang#

ECCV 2024 Oral Presentation

First zero-shot 360° super-resolution leveraging Stable Diffusion priors via tangent-plane interaction and gradient-based decomposition, no training needed.

Paper | Project Page | GitHub

CVPRW 2023

OPDN: Omnidirectional Position-Aware Deformable Network for Omnidirectional Image Super-Resolution

Xiaopeng Sun*, Weiqi Li*, Zhenyu Zhang, Qiufang Ma, Xuhan Sheng, et al.

CVPR Workshops 2023 NTIRE 2023 360° Panoramic SR Challenge — Champion

Position-aware deformable network tackling ERP geometric distortions; champion of the NTIRE 2023 360° panoramic SR challenge.

Paper

Video Understanding

PRCV 2025

Bridging the Point to Boundary Gap for Point-supervised Temporal Action Localization with Single-stage Inference

Junshi Yang, Shenglan Liu, Xuhan Sheng, et al.

PRCV 2025

Hierarchical point-to-boundary framework converting sparse point labels into boundary-accurate pseudo labels via Gaussian-prior boundary enhancement.

Paper

🎓 Educations

Institute of Science Tokyo, Department of Computer Science

2026.10 - Present · Incoming Ph.D. Student

Advisor: Prof. Rio Yokota

Peking University, VILLA Lab

2023.09 - 2026.06 · Master Student

Advisor: Prof. Jian ZHANG; GPA: 3.87/4.00

Dalian University of Technology

2019.09 - 2023.06 · Bachelor Student

Advisor: Prof. Xu JIA; GPA: 4.17/5.00

🏅 Honors and Awards

• 2024, Peking University Academic Excellence Award.
• 2023, Dalian University of Technology University-level Outstanding Graduate.
• 2022, Dalian University of Technology First-class Academic Excellence Scholarship.
• 2021, Dalian University of Technology Second-class Academic Excellence Scholarship.
• 2020, Dalian University of Technology First-class Academic Excellence Scholarship.

😊 Hobbies

Click image to zoom.

I love playing flutes (Irish flute, Boehm flute, Shakuhachi). I love Hatsune Miku. I am learning Cantonese.