Xiefan Guo
Portrait of Xiefan Guo
Email: guoxiefan@gmail.com
Google Scholar
GitHub
Twitter

I am a Ph.D. student in Computer Science at Beihang University (BUAA),
where I am advised by Prof. Di Huang.

My research interests are in computer vision and generative models, especially diffusion-based image and video generation.

I received my M.S. from Beihang University in 2023 and my B.E. from Tianjin University in 2020. I have interned at Shanghai AI Laboratory, Alibaba TongYi Lab, and Alibaba DAMO Academy.

Open to Opportunities:I expect to graduate in December 2026 and am currently on the job market for research positions in academia and industry, particularly in image/video generation and world models.

News

  • Jun 2026
    Our paper, titled “EruDiff: Refactoring Knowledge in Diffusion Models for Advanced Text-to-Image Synthesis” is accepted by ECCV 2026.
  • Jun 2026
    Our paper, titled “G3AFT: Glance Guided Gradient Aligned Fine-Tuning for Visual Autoregressive Models” is accepted by ECCV 2026.
  • May 2026
    Our paper, titled “What Makes Synthetic Data Effective in Image Segmentation” is accepted by ICML 2026.
  • Feb 2026
    Our paper, titled “CTCal: Rethinking Text-to-Image Diffusion Models via Cross-Timestep Self-Calibration” is accepted by CVPR 2026.

Preprints

  1. arXiv'24

    I4VGen: Image as Free Stepping Stone for Text-to-Video Generation.

    Xiefan Guo, Jinlin Liu, Miaomiao Cui, Liefeng Bo, and Di Huang.
    Paper / arXiv / Code / Project

Selected Publications Full list on Google Scholar.

  1. ECCV'26

    EruDiff: Refactoring Knowledge in Diffusion Models for Advanced Text-to-Image Synthesis.

    Xiefan Guo, Xinzhu Ma, Haoxiang Ma, Zihao Zhou, and Di Huang.
    Paper / arXiv / Code
  2. ECCV'26

    G3AFT: Glance Guided Gradient Aligned Fine-Tuning for Visual Autoregressive Models.

    Jiayi Zhang, Xiefan Guo, Xinzhu Ma, and Di Huang.
    Paper / Code
  3. ICML'26

    What Makes Synthetic Data Effective in Image Segmentation.

    Jinjin Zhang, Xiefan Guo, Yizhou Jin, Nan Zhou, and Di Huang.
    Paper / arXiv / Code
  4. CVPR'26

    CTCal: Rethinking Text-to-Image Diffusion Models via Cross-Timestep Self-Calibration.

    Xiefan Guo, Xinzhu Ma, Haiyu Zhang, and Di Huang.
    Paper / arXiv / Code
  5. ICCV'25

    ShortFT: Diffusion Model Alignment via Shortcut-based Fine-Tuning.

    Xiefan Guo, Miaomiao Cui, Liefeng Bo, and Di Huang.
    Paper / arXiv / Code / Project / Poster
  6. CVPR'25

    Diffusion-4K: Ultra-High-Resolution Image Synthesis with Latent Diffusion Models.

    Jinjin Zhang, Qiuyu Huang, Junjie Liu, Xiefan Guo, and Di Huang.
    Paper / arXiv / Code / Dataset
  7. CVPR'24

    InitNO: Boosting Text-to-Image Diffusion Models via Initial Noise Optimization.

    Xiefan Guo, Jinlin Liu, Miaomiao Cui, Jiankai Li, Hongyu Yang, and Di Huang.
    Paper / arXiv / Code / Project / Slide / Poster
  8. CVPR'24

    Leveraging Predicate and Triplet Learning for Scene Graph Generation.

    Jiankai Li, Yunhong Wang, Xiefan Guo, Ruijie Yang, Weixin Li.
    Paper / arXiv / Code
  9. CVPR'22

    ABPN: Adaptive Blend Pyramid Network for Real-Time Local Retouching of Ultra High-Resolution Photo.

    Biwen Lei, Xiefan Guo, Hongyu Yang, Miaomiao Cui, Xuansong Xie, and Di Huang.
    Paper / Dataset
  10. ICCV'21

    Image Inpainting via Conditional Texture and Structure Dual Generation.

    Xiefan Guo, Hongyu Yang, and Di Huang.
    Paper / arXiv / Code / Slide / Poster

Professional Activities

Conference ReviewerCVPR, ICCV, ECCV, ICML, NeurIPS, ICLR, SIGGRAPH, SIGGRAPH Asia, AAAI, AISTATS, ACM MM, ICMR, ECAI, BMVC, PRCV.

Journal ReviewerTPAMI, IJCV, TVCG, TMM, TCSVT, CVIU, FCS, J-STSP.