About me

My name is Junhao Cheng (程钧豪), currently an MPhil student at CityUHK. I received my bachelor's degree from Sun Yat-sen University (SYSU) in 2025, where I was fortunate to be supervised by Prof. Xiaodan Liang. And I spent a wonderful year as a research intern at Tencent ARC Lab. I also had the opportunity to work with Prof. Ming-Hsuan Yang.

I am currently a research intern at the Kuaishou Kling team. My research interests lie in Interactive AI, particularly foundation models and novel applications for video reasoning and generation.

I am currently seeking Industry Roles or Startup Opportunities (graduating Fall 2027). I’d be happy to connect via WeChat or Email.

News

2026.06 🎉🎉 One paper is accepted by ECCV 2026.
2026.05 Release VLM-as-Teacher, enhancing video generation model with reasoning ability.
2026.02 🎉🎉 One paper is accepted by CVPR 2026, along with two at AISTORY Workshop.
2025.11 Release Video-as-Answer, extending next-event prediction to video answer.
2025.06 🎉🎉 One paper is accepted by ICCV 2025.
2025.06 Release Video-Holmes, evaluating MLLMs for complex video reasoning.
2025.04 Release AnimeGamer (300+Stars✨), transforming characters from anime films into interactive entities.
2024.06 Release AutoStudio (400+Stars✨), generating comic book with multi-character consistency.
2024.05 🎉🎉 One paper is accepted by ACL 2024.

Education

MPhil @ City University of Hong Kong

2025 - 2027 (expected)

Supervisor: Jing Liao

Undergrad @ Sun Yat-sen University

2021 - 2025

Supervisor: Xiaodan Liang

Internships

Research Intern @ Kuaishou, Kling Team

2025.07 - Present

Mentor: Liang Hou, Xin Tao

Research Intern @ Tencent, ARC Lab

2024.09 - 2025.06

Mentor: Yuying Ge

Research Intern @ Lenovo Research

2023.09 - 2024.08

Research Intern @ CIBR

2023.03 - 2023.08

Mentor: Yunzhe Liu

Selected Publications

VANS

Video-as-Answer: Predict and Generate Next Video Event with Joint-GRPO

Junhao Cheng, Liang Hou, Xin Tao, Jing Liao

CVPR 2026
Paper Code
AnimeGamer

AnimeGamer: Infinite Anime Life Simulation with Next Game State Prediction

Junhao Cheng, Yuying Ge, Yixiao Ge, Jing Liao, Ying Shan

ICCV 2025
Paper Code
Video-Holmes

Video-Holmes: Can MLLM Think Like Holmes for Complex Video Reasoning?

Junhao Cheng, Yuying Ge, Teng Wang, Yixiao Ge, Jing Liao, Ying Shan

ECCV 2026
Paper Code
BD-Diff

BD-Diff: Generative Diffusion Model for Image Deblurring on Unknown Domains

Junhao Cheng, Wei-Ting Chen, Xi Lu, Ming-Hsuan Yang

ArXiv 2025
Paper
AutoStudio

AutoStudio: Crafting Consistent Subjects in Multi-turn Interactive Image Generation

Junhao Cheng, Xi Lu, Hanhui Li, Khun Loun Zai, et al.

CVPRW 2026
Paper Code
Isolate

Object Isolated Attention for Consistent Story Visualization

Xiangyang Luo, Junhao Cheng, Yifan Xie, Xin Zhang, et al.

ICME 2025
Paper
VisDiaHalBench

VisDiaHalBench: A Visual Dialogue Benchmark For Diagnosing Hallucination in Large Vision-Language Models

Qingxing Cao, Junhao Cheng, Xiaodan Liang, Liang Lin

ACL 2024
Paper Code
DKFormer

Integrating Domain Knowledge into Transformer for Short-Term Wind Power Forecasting

Junhao Cheng, Xing Luo, Zhi Jin

Energy (JCR-Q1)
Paper