Welcome to my homepage! I am Xiaoyang Liu (刘潇阳), a second-year M.S. student in Computer Science at Shanghai Jiao Tong University. Currently, I am collaborating closely with Prof. Yulun Zhang in his research group. My previous research has focused on Image Restoration, Generative Models, and Model Compression. My current research interests lie in multimodal foundation models, particularly VLM architectures, visual representation learning, and cross-modal interaction for visual understanding and reasoning. I am especially interested in how visual information is encoded, integrated, and utilized throughout multimodal reasoning. I am actively seeking research internship and collaboration opportunities in these directions. Please feel free to contact me.

🔥 News

📝 Publications

* denotes Equal Contribution, $\dagger$ denotes Corresponding Author.

NeurIPS 2026

PRISM: Prior Rectification and Uncertainty-Aware Structure Modeling for Diffusion-Based Text Image Super-Resolution

Zihang Xu*, Xiaoyang Liu*, Zheng Chen, Yulun Zhang$\dagger$, Xiaokang Yang

In NeurIPS 2026 | Paper | Code

ECCV 2026

Freqformer: Image-Demoiréing Transformer via Effective Frequency Decomposition

Xiaoyang Liu*, Bolin Qiu*, Zheng Chen, Libo Zhu, Zihan Zhou, Kai Liu, Jiezhang Cao, Yulun Zhang$\dagger$

In ECCV 2026 | Paper | Code

ICLR 2026

FideDiff: Efficient Diffusion Model for High-Fidelity Image Motion Deblurring

Xiaoyang Liu*, Zhengyan Zhou*, Zihang Xu, Jiezhang Cao, Zheng Chen, Yulun Zhang$\dagger$

In ICLR 2026 | Paper | Code | OpenReview

AAAI 2025

Interacted Object Grounding in Spatio-Temporal Human-Object Interactions

Xiaoyang Liu*, Boran Wen*, Xinpeng Liu*, Zizheng Zhou, Hongwei Fan, Cewu Lu, Lizhuang Ma, Yulong Chen$\dagger$, Yong-Lu Li$\dagger$

In AAAI 2025 | Paper | Code | Huggingface

NTIRE Workshop

• “The Fourth Challenge on Image Super-Resolution ($\times$4) at NTIRE 2026: Benchmark Results and Method Overview”. Zheng Chen et al. In CVPR 2026 Workshops (NTIRE). [Paper]

• “The First Challenge on Mobile Real-World Image Super-Resolution at NTIRE 2026: Benchmark Results and Method Overview”. Jiatong Li et al. In CVPR 2026 Workshops (NTIRE). [Paper]

• “The First Challenge on Remote Sensing Infrared Image Super-Resolution at NTIRE 2026: Benchmark Results and Method Overview”. Kai Liu et al. In CVPR 2026 Workshops (NTIRE). [Paper]

Image Restoration

• “One-Step Diffusion Model for Image Motion-Deblurring”. Xiaoyang Liu, Yuquan Wang, Zheng Chen, Jiezhang Cao, He Zhang, Yulun Zhang$\dagger$, Xiaokang Yang. arXiv preprint arXiv:2503.06537. [Paper] [Project]

• “RIFLE: Removal of Image Flicker-Banding via Latent Diffusion Enhancement”. Libo Zhu*, Zihan Zhou*, Xiaoyang Liu, Weihang Zhang, Keyu Shi, Yifan Fu, Yulun Zhang$\dagger$. NeurIPS 2026. [Paper] [Project]

Model Compression

• “QuantDemoire: Quantization with Outlier Aware for Image Demoiréing”. Zheng Chen*, Kewei Zhang*, Xiaoyang Liu, Weihang Zhang, Mengfan Wang, Yifan Fu, Yulun Zhang$\dagger$. NeurIPS 2026. [Paper] [Project]

• “BinaryDemoire: Moiré-Aware Binarization for Image Demoiréing”. Zheng Chen*, Zhi Yang*, Xiaoyang Liu, Weihang Zhang, Mengfan Wang, Yifan Fu, Linghe Kong, Yulun Zhang$\dagger$. arXiv preprint arXiv:2602.03176. [Paper] [Project]

Human-Object Interaction

• “DJ-RN++: Detecting and Recovering Human-Object Interactions from Still Images”. Yong-Lu Li*, Xinpeng Liu*, Boran Wen, Xiaoyang Liu, Cewu Lu$\dagger$. Under Review.

🎖 Honors and Awards

• National Scholarship (Top 1%, Graduate Level), 2026
• National Scholarship (Top 1%, Graduate Level), 2025
• First-Class Academic Scholarship for 2025 Master’s Students, SJTU, 2025
• Ye Jun & Shen Nanpeng Zhiyuan Outstanding Scholarship (Top 1%), SJTU, 2025
• Outstanding Graduate of SJTU, 2025
• Outstanding Bachelor Thesis (Top 1%), SJTU, 2025
• Liu Yongling Scholarship (Top 5%), SJTU, 2024
• Huawei Scholarship (Top 5%), SJTU, 2023
• Zhiyuan Honors Scholarship (Top 5%), SJTU, 2021, 2022, 2023, 2024
• Merit Scholarship (Top 10%), SJTU, 2022, 2023, 2024
• Merit Student (Top 5%), SJTU, 2022
• First Prize, “Internet + Rural Revitalization” Social Practice, SJTU, 2022

📖 Education

• 2025.09 - 2028.03 (expected), M.S. in Computer Science, Shanghai Jiao Tong University.
• 2021.09 - 2025.06, B.S. in Computer Science and Zhiyuan Honors Degree, Shanghai Jiao Tong University.
  • Ranked 11/103, with A+ in 21 courses and A in 30 courses out of 67 courses.
  • Zhiyuan Honors Degree is an elite program for top 5% talented students.

💻 Internships

Huawei, AI Algorithm Intern, 2026.07 - 2026.09.

• Trained VLM understanding and editing-instruction generation modules; performed multimodal SFT on a 35B VLM with LLaMA-Factory, covering multi-source image-text data integration, LoRA fine-tuning, distributed training, and offline inference evaluation.
• Customized the GRPO training and reward pipeline with verl: sampled multiple editing instructions per image in parallel, generated candidate results with an image-to-image model, and optimized the policy using result-level scores fused from pose and composition reward models to align editing instructions with final image quality.
• Built asynchronous resource orchestration across VLMs, image-editing models, and reward models with Ray, aiohttp, vLLM, and FSDP2; developed an end-to-end optimization pipeline for content hallucination, abnormal human poses, and composition imbalance. Training was conducted on a four-node Huawei cluster with 32 Ascend 910B NPUs in total.

💼 Collaborative Projects

Exploratory Collaboration on Image/Video Demoiréing and Flicker-Banding Removal (Huawei). Project Lead. Responsible for task investigation, new dataset construction, image/video demoiréing network design, project coordination, and delivery.
Official documentation confirms the deployment of our technology on Huawei Pura 90 series smartphones.

University-Industry Research on Snapshot Image Quality Enhancement (Xiaomi). Project Lead. Responsible for project planning, coordination, and delivery for image motion deblurring and motion artifact removal.

Enhancing Reasoning Capabilities of Lightweight VLMs (Huawei). Project Member. Responsible for surveying VLM architectures and distillation methods, and conducting preliminary distillation experiments.

Acceleration and Quantization of Multimodal Human Video Generation Models (Mango TV). Project Member. Responsible for deploying caching mechanisms for Alibaba Tongyi Wanxiang video generation models WAN2.1 and WAN2.2, achieving a 3x speedup.

Mobile Customization of Efficient Diffusion-Based Image Reconstruction Algorithms (Transsion). Project Member. Responsible for the image motion deblurring task.

🤝 Service

• Academic Service
  • Reviewer: PRCV 2025, ICCV 2025, NeurIPS 2025, CVPR 2026, ECCV 2026, ACM MM 2026, ICLR 2027
  • Volunteer, 1st China Embodied Intelligence Conference, 2024.03
• Student Leadership
  • Class Director, 2025 Cohort Zhiyuan Engineering Honors Program, Zhiyuan College, SJTU, 2025.08 - Present
  • Publicity Officer, CS25M031 Class, SJTU, 2025.09 - Present
  • Vice Monitor, F2103302 Class, SJTU, 2021.11 - 2025.06
  • Student Union Officer, SEIEE, SJTU, 2022.03 - 2023.03
• Community Service
  • Volunteer, Shanghai International Marathon, 2022, 2023
  • Volunteer, Shanghai West Bund Art Museum, 2024