Tianhe Wu

Ph.D. Student in City University of Hong Kong
Research Intern in Qwen Team
E-mail: tianhewu-c@my.cityu.edu.hk
Google Scholar / Github

About Me

Hi, I'm Tianhe Wu, a Ph.D. student at the CityU, where I am supervised by Prof. Kede Ma. My current research focuses on AIGC and AGI, with a particular interest in diffusion distillation and data valuation. I am a serious researcher and always happy to connect—feel free to reach out to me via email for discussion and collaboration.

Before joining CityU, I completed my M.Sc. degree (2022-2025) at Tsinghua University, under the supervision of Prof. Yujiu Yang. I received my B.Eng. degree (2018-2022) from the Beijing University of Technology.

Internship

	Qwen Team Research Intern, focusing on Generative Modeling Supervisor: Chenfei Wu	Apr. 2026 - Present
	YLab, OPPO Research Institute Research Intern, focusing on VLM Supervisor: Prof. Lei Zhang	Apr. 2024 - Apr. 2026
	Department of Computer Science, City University of Hong Kong Research Assistant, focusing on VLM Supervisor: Prof. Kede Ma	Dec. 2023 - Aug. 2025

News

2025-09: Paper VisualQuality-R1 obtained NeurIPS 2025 (my first main conference Spotlight paper), CityU and OPPO.
2025-02: One paper was accepted at CVPR 2025, CityU and OPPO.
2024-07: One paper was accepted at ECCV 2024, Tsinghua and CityU.
2023-09: A paper was accepted to NeurIPS 2023, Tsinghua University.
2022-06: I published my first conference paper, MANIQA at CVPR Workshopss 2022 (Oral), Tsinghua University.
2022-04: Our team at Tsinghua won the NTIRE 2022 Perceptual Image Quality Assessment Challenge, taking first place in both Track 1 (Full-Reference) and Track 2 (No-Reference).

Representative Publications

Preprint 2025 Diversity-Preserved Distribution Matching Distillation for Fast Visual Synthesis
Tianhe Wu*, Ruibin Li*, Lei Zhang, Kede Ma

arXiv / Code

A simple yet effective approach to preserving sample diversity under DMD, with no perceptual backbone, no discriminator, no auxiliary networks, and no additional ground-truth images.

NeurIPS 2025 Spotlight VisualQuality-R1: Reasoning-Induced Image Quality Assessment via Reinforcement Learning to Rank
Tianhe Wu, Jian Zou, Jie Liang, Lei Zhang, Kede Ma

arXiv / Code

A reasoning-enabled visual quality model capable of delivering both qualitative descriptions and quantitative ratings.

CVPR Workshops 2022 Oral MANIQA: Multi-dimension Attention Network for No-Reference Image Quality Assessment
Sidi Yang*, Tianhe Wu*, Shuwei Shi, Shanshan Lao, Yuan Gong, Mingdeng Cao, Jiahao Wang, Yujiu Yang

arXiv / Code

The champion method for NTIRE2022 NR IQA.

Other Publications

	ACL Findings 2026 Beyond Length Scaling: Synergizing Breadth and Depth for Generative Reward Models Qiyuan Zhang, Yufei Wang, Tianhe Wu, Can Xu, Qingfeng Sun, Kai Zheng, Xue Liu, Chen Ma arXiv / Code Test time scaling for generative reward models.
	NeurIPS 2025 DP2O-SR: Direct Perceptual Preference Optimization for Real-World Image Super-Resolution Rongyuan Wu, Lingchen Sun, Zhengqiang Zhang, Shihao Wang, Tianhe Wu, Qiaosi Yi, Shuai Li, Lei Zhang arXiv / Code Direct perceptual preference optimization for image super-resolution.
	Preprint 2025 The Unanticipated Asymmetry Between Perceptual Optimization and Assessment Jiabei Zhang, Qi Wang, Siyu Wu, Du Chen, Tianhe Wu^† arXiv / Code Our comprehensive study highlights the asymmetries between assessment and optimization.
	CVPR 2025 Toward Generalized Image Quality Assessment: Relaxing the Perfect Reference Quality Assumption Du Chen, Tianhe Wu, Kede Ma, Lei Zhang arXiv / Project Page / Code / Dataset / Benchmark A generalized FR-IQA model for low-level vision, applicable to both perfect and imperfect reference scenarios.
	ECCV 2024 A Comprehensive Study of Multimodal Large Language Models for Image Quality Assessment Tianhe Wu, Kede Ma, Jie Liang, Yujiu Yang, Lei Zhang arXiv / Code A systematic study on the ability of multimodal large language models to perceive and assess image quality.
	NeurIPS 2023 Assessor360: Multi-sequence Network for Blind Omnidirectional Image Quality Assessment Tianhe Wu, Shuwei Shi, Haoming Cai, Mingdeng Cao, Jing Xiao, Yinqiang Zheng, Yujiu Yang arXiv / Code We present a new framework for blind omnidirectional IQA.
	CVPR Workshops 2022 Oral Attentions Help CNNs See Better: Attention-based Hybrid Image Quality Assessment Network Shanshan Lao, Yuan Gong, Shuwei Shi, Sidi Yang, Tianhe Wu, Jiahao Wang, Weihao Xia, Yujiu Yang arXiv / Code The champion method for NTIRE2022 FR IQA.

Education

	City University of Hong Kong Ph.D. Student in Department of Computer Science Supervisor: Prof. Kede Ma	Sep. 2025 - Present
	Tsinghua University Master Degree in Big Data Technology and Engineering Supervisor: Prof. Yujiu Yang	Sep. 2022 - Jun. 2025
	Beijing University of Technology Bachelor Degree in Computer Science	Sep. 2018 - Jun. 2022