Xuanhua He — Agentic World Models

About Me

I am currently a Ph.D. student in Computer Science and Engineering at HKUST, advised by Prof. Qifeng Chen. Previously, I obtained my M.S. from USTC, working with Prof. Jie Zhang and Chengjun Xie. I received my B.Eng. degree from Xiamen University in 2022, where I conducted research at the MOCOM Lab under the guidance of Prof. Yongxuan Lai.

My research sits at the intersection of agentic world models, video intelligence, and self-improving agents. I am interested in systems that perceive dynamic worlds, take purposeful actions, verify their outcomes, and turn experience into better future behavior.

01 · MODEL

Understand the world

Learn temporal, spatial, and physical structure from open-world video.

02 · ACT

Reason with tools

Ground decisions through search, interaction, and closed-loop feedback.

03 · EVOLVE

Improve from experience

Use verification and self-distillation to make long-horizon agents stronger.

Current question: Can a world model become the environment, memory, and verifier an agent needs to recursively improve?

News

Aug 2026 Two papers accepted to EMNLP 2026 on self-evolving search agents and video deep research.

Jun 2026 One paper accepted to ECCV 2026.

Apr 2026 One paper accepted to ICML 2026.

Feb 2026 Two papers accepted to CVPR 2026.

Jan 2026 One paper accepted to ICLR 2026.

Nov 2025 Four papers accepted to AAAI 2026.

Sep 2025 Started my Ph.D. journey at HKUST.

Sep 2025 Joined Meituan as a Research Intern.

May 2025 Released two video editing papers on arXiv: FullDiT2 and UNIC.

Feb 2025 Joined Kuaishou Kling AI as a Research Intern.

Feb 2025 GameGen-X accepted to ICLR 2025!

Oct 2024 Released GameGen-X, an interactive open-world game video generation model.

Oct 2024 Completed research internship at Tencent.

May 2024 Released ID-Animator, a zero-shot identity-preserving video generation model.

Apr 2024 Joined Tencent as a Research Intern.

Experience

Sep 2025

Present

LongCat Avatar Team, Meituan

Research Intern, Talented Plan · Video Avatar and World Model · Mentor: Yong Zhang

Feb 2025

Aug 2025

Kling AI Team, Kuaishou

Research Intern · Video Editing · Mentor: Quande Liu

Apr 2024

Oct 2024

LightSpeed, Tencent

Research Intern · Interactive Video Generation · Mentor: Quande Liu

Nov 2023

Mar 2024

Baidu

CV Engineer Intern · Video Generation

Jun 2023

Nov 2023

Horizon Robotics

Research Intern · Low-level Vision · Mentor: Guoli Wang

Selected Publications

Self-Improving & Multimodal Agents

Verify · Reflect · Evolve DeepSearch World

DeepSearch-World: Self-Distillation for Deep Search Agents in a Verifiable Environment

Xinyu Geng, Xuanhua He*(Co-First Author), Sixiang Chen, Yanjing Xiao, Fan Zhang, Shijue Huang, Haitao Mi, Zhenwen Liang, Tianqing Fang, Yi R. Fung

EMNLP 2026

Paper Code & Env

Ground · Search · Reason Video Searcher

VideoSearcher: Empowering Video Deep Research with Multi-Tool Agentic Reasoning via Reinforcement Learning

Zhenkun Gao*, Yicheng Bao*, Jinlong Peng*, Xueheng Li*, Suyuan Huang*, Bangwei Liu, Kunquan Li, Zhenye Gan, Tao Hu, Chengjun Xie, Xuanhua He✉, Zhizhong Zhang, Xin Tan, Chengjie Wang, Yuan Xie

EMNLP 2026

Project Paper Code

Canvas · Memory · Action Jarvis Hub

JarvisHub: An Open Harness for Canvas-Native Multimodal Creative Agents

Yunlong Lin, Zixu Lin, Zhaohu Xing, Biqiang Li, Chenxin Li, Haonan Wang, ..., Xuanhua He, ..., Tianyu Pang, Xiangyu Yue

ArXiv 2026

Project Paper Code

Probe · Zoom · Judge Q-Probe

Q-Probe: Scaling Image Quality Assessment to High Resolution via Context-Aware Agentic Probing

Xiang Li, Xueheng Li, Yu Wang, Xuanhua He✉(Corresponding Author), Zhangchi Hu, Weiwei Yu, Chengjun Xie

ArXiv 2026

Paper

Interactive Video & World Models

ICDepth: Taming Video Diffusion Models for Video Depth Estimation via In-Context Conditioning

Xuanhua He, Jiaxin Xie, Mingzhe Zheng, Qifeng Chen

ECCV 2026

Project PDF

Infinite-World: Scaling Interactive World Models to 1000-Frame Horizons via Pose-Free Hierarchical Memory

Ruiqi Wu, Xuanhua He(Co-first author), Meng Cheng, Tianyu Yang, Yong Zhang, Zhuoliang Kang, Xunliang Cai, Xiaoming Wei, Chunle Guo, Chongyi Li, Ming-Ming Cheng

ICML 2026

Project PDF Code

Active Intelligence in Video Avatars via Closed-loop World Modeling

Xuanhua He, Tianyu Yang, Ke Cao, Ruiqi Wu, Cheng Meng, Yong Zhang, Zhuoliang Kang, Xiaoming Wei, Qifeng Chen

CVPR 2026

Project PDF Code

ContextFlow: Training-Free Video Object Editing via Adaptive Context Enrichment

Yiyang Chen, Xuanhua He*(corresponding author), Xiujun Ma, Yue Ma

AAAI 2026

Project PDF Code

FullDiT2: Efficient In-Context Conditioning for Video Diffusion Transformers

Xuanhua He, Quande Liu, Zixuan Ye, Weicai Ye, Qiulin Wang, Xintao Wang, Qifeng Chen, Pengfei Wan, Di Zhang, Kun Gai

Arxiv 2025

Project PDF

UNIC: Unified In-Context Video Editing

Zixuan Ye, Xuanhua He*(Co-First Author), Quande Liu, Qiulin Wang, Xintao Wang, Pengfei Wan, Di Zhang, Kun Gai, Qifeng Chen, Wenhan Luo

ICLR 2026

Project PDF

GameGen-X: Interactive Open-World Game Video Generation

Haoxuan Che, Xuanhua He*(Co-First Author), Quande Liu, Cheng Jin, Hao Chen

ICLR 2025

Project PDF Code

ID-Animator: Zero-Shot Identity-Preserving Human Video Generation

Xuanhua He, Quande Liu, Shengju Qian, Xin Wang, Tao Hu, Ke Cao, Keyu Yan, Jie Zhang

ArXiv 2024

Project PDF Code

Low-level Vision

Cross-Scale · Benchmark ScaleFormer

Cross-Scale Pansharpening via ScaleFormer and the PanScale Benchmark

Ke Cao, Xuanhua He*(Co-First Author, Project Leader), Xueheng Li, Lingting Zhu, Yingying Wang, Ao Ma, Zhanjie Zhang, Man Zhou, Chengjun Xie, Jie Zhang

CVPR 2026

PDF Code Dataset

Pyramid · Frequency Dual Domain Injection

Pyramid Dual Domain Injection Network for Pan-sharpening

Xuanhua He, Keyu Yan, Rui Li, Chengjun Xie, Jie Zhang, Man Zhou

ICCV 2023

Paper Code

Frequency MoE FAME-Net

Frequency-adaptive Pan-sharpening with Mixture-of-Experts

Xuanhua He, Keyu Yan, Rui Li, Chengjun Xie, Jie Zhang, Man Zhou

AAAI 2024

PDF Code

RAW → sRGB FourierISP

Enhancing RAW-to-sRGB with Decoupled Style Structure in Fourier Domain

Xuanhua He, Tao Hu, Guoli Wang, Zejin Wang, Run Wang, Qian Zhang, Keyu Yan, Ziyi Chen, Rui Li, Chengjun Xie, Jie Zhang, Man Zhou

AAAI 2024

PDF Code

Distilled Fusion DTPF

Distilling Textual Priors from LLM to Efficient Image Fusion

Ran Zhang, Xuanhua He* (Project Leader), Ke Cao, Liu Liu, Li Zhang, Man Zhou, Jie Zhang, Meng Wang

IEEE TCSVT 2025

PDF Code

Awards

Outstanding Graduate, USTC (2025)
National Scholarship, USTC (2023, 2024)
Yuequn Scholarship for Academic Excellence, USTC (2023)