Jinfan Hu

Ph.D. · Computer Applied Technology

Jinfan Hu

Ph.D. in Computer Applied Technology @ SIAT, CAS | Advised by Prof. Chao Dong
Low-level Vision, Model Interpretability, and Generative Models.

Low-level Vision Interpretability Generative Models
01

Biography

Jinfan Hu received his Ph.D. degree in Computer Applied Technology from XPixel Group, Shenzhen Institutes of Advanced Technology (SIAT), Chinese Academy of Sciences (CAS), advised by Prof. Chao Dong (Sep. 2022 – Jun. 2026). He received his M.S. degree in Mathematics in 2022, supervised by Prof. Ting-Zhu Huang and Prof. Liang-Jian Deng, and his B.S. degree in Information and Computing Science in 2019 from the University of Electronic Science and Technology of China (UESTC), Chengdu, China. His research interests include low-level vision (image and video restoration, super-resolution, and generative models) and model interpretability (causal analysis in deep networks and model generalization).

02

Headlines

09/2026

Two papers are accepted by NeurIPS

The position paper Evaluation of Visual Processing Should Be Human-Centered, Not Metric-Centered (equal contribution) argues that visual-processing evaluation should follow human perception, rather than a single image-quality metric.

Harnessing Image Diffusion Prior for Photo-Realistic Video Restoration transfers image diffusion priors to photo-realistic video restoration.

03/2026

One paper is published in Pattern Recognition

This paper demonstrates that the common strategy of blindly expanding the training set is ineffective. Instead, the key to better generalization lies in guiding the network to learn the image content rather than the degradation!

Takeaway: Teach networks to see the content, not just remove degradation.
Generalization Comparison
08/2025

We propose HYPIR: A Step Toward the Next Generation of Image Restoration

HYPIR introduces a simple yet powerful paradigm: fine-tuning a pre-trained diffusion model with adversarial (GAN) loss — no diffusion sampling, no extra adapters. This achieves an unprecedented balance of speed, fidelity, and quality. Initializing GAN training from a pre-trained diffusion model anchors the model near the natural image manifold, ensuring stable training, faster convergence, and superior results — all in a single forward pass.

Official Webpage: The high-speed, high-performance image restoration model is now available on 明犀AI and SupPixel AI.

HYPIR Teaser
03/2025

One paper is accepted by TPAMI

We propose a model/task-agnostic interpreting method (CEM) for low-level vision models, bringing causality analysis into the field. We hope this work contributes to a deeper understanding and enhancement of low-level vision models!

Key Insight: Correlation does not imply causation — only causal analysis reveals why.
CEM Teaser
02/2025

The book 《底层视觉之美》 has been published

Co-authored with my supervisor Prof. Chao Dong. We hope this book provides valuable insights for researchers and enthusiasts in the Low-level Vision field!

Available on: JD.com

The Beauty of Low-level Vision Book
01/2024

We release a groundbreaking image restoration method (SUPIR)

That harnesses the generative prior and the power of model scaling up. Our method demonstrates unprecedented performance in real-world image restoration tasks!

Experience it now: 明犀AI and SupPixel AI.

SUPIR Teaser
03

Publications

Book
Selected Papers

J. Hu*, Z. You*, J. Gu, K. Zhu, T. Xue, C. Dong. Revisiting the Generalization Problem of Low-level Vision Models Through the Lens of Image Deraining. Pattern Recognition, 2026.

PR · 2026

J. Hu, J. Gu, S. Yu, F. Yu, Z. Li, Z. You, C. Lu, C. Dong. Interpreting Low-level Vision Models with Causal Effect Maps. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2025.

TPAMI · 2025

J.-F. Hu, T.-Z. Huang, L.-J. Deng, H.-X. Dou, D. Hong, and G. Vivone. Fusformer: A Transformer-based Fusion Approach for Hyperspectral Image Super-resolution. IEEE Geoscience and Remote Sensing Letters (GRSL), 2022. ESI Highly Cited

GRSL · 2022

J.-F. Hu, T.-Z. Huang, L.-J. Deng, T.-X. Jiang, G. Vivone, and J. Chanussot. Hyperspectral Image Super-Resolution via Deep Spatiospectral Attention Convolutional Neural Networks. IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2021. ESI Highly Cited

TNNLS · 2021

F. Yu*, J. Hu*, Z. You, X. Yin, H. An, X. Lin, H. Li, C. Dong, J. Gu. Evaluation of Visual Processing Should Be Human-Centered, Not Metric-Centered. Advances in Neural Information Processing Systems (NeurIPS, Position Paper Track), 2026.

NeurIPS · 2026

F. Yu, H. An, J. Hu, X. Lin, Z. You, J. Wang, H. Li, C. Dong, J. Gu. Harnessing Image Diffusion Prior for Photo-Realistic Video Restoration. Advances in Neural Information Processing Systems (NeurIPS), 2026.

NeurIPS · 2026

X. Yin, J. Hu, Z. You, K. Yan, Y. Tang, C. Dong, J. Gu. How Far Have We Gone in Generative Image Restoration? A Study on its Capability, Limitations and Evaluation Practices. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR, Findings Track), 2026.

CVPR Findings · 2026

X. Lin, F. Yu, J. Hu, Z. You, W. Shi, Jimmy S. Ren, J. Gu, C. Dong. Harnessing Diffusion-Yielded Score Priors for Image Restoration. ACM SIGGRAPH Asia, 2025.

SIGGRAPH Asia · 2025

F. Yu, J. Gu, J. Hu, Z. Li, and C. Dong. UniCon: Unidirectional Information Flow for Effective Control of Large-Scale Diffusion Models. International Conference on Learning Representations (ICLR), 2025.

ICLR · 2025

F. Yu, J. Gu, Z. Li, J. Hu, X. Kong, X. Wang, J. He, Y. Qiao, and C. Dong. Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic Image Restoration In the Wild. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024.

CVPR · 2024

R. Ran, L.-J. Deng, T.-X. Jiang, J.-F. Hu, J. Chanussot, and G. Vivone. GuidedNet: A General CNN Fusion Framework via High-resolution Guidance for Hyperspectral Image Super-resolution. IEEE Transactions on Cybernetics (TCYB), 2023. ESI Highly Cited

TCYB · 2023

X. Liu*, J. Hu*, X. Chen, and C. Dong. UDC-UNet: Under-Display Camera Image Restoration via U-shape Dynamic Network. European Conference on Computer Vision Workshop (ECCVW), 2022.

ECCVW · 2022

S. Peng, L.-J. Deng, J.-F. Hu, and Y.-W. Zhuo. Source-Adaptive Discriminative Kernels based Network for Remote Sensing Pansharpening. International Joint Conferences on Artificial Intelligence (IJCAI), 2022.

IJCAI · 2022

(* Equal contributions)

04

Products

For a comprehensive exploration of our technologies, visit our official websites: 明犀AI and SupPixel AI.

05

Experiences

09/2022-06/2026
Ph.D. in Computer Applied Technology. (Advisor: Prof. Chao Dong); XPixel Group, Shenzhen Institutes of Advanced Technology (SIAT), Chinese Academy of Sciences (CAS)
09/2019-06/2022
M.S. in Mathematics. (Advisors: Prof. Ting-Zhu Huang and Prof. Liang-Jian Deng); University of Electronic Science and Technology of China (UESTC)
09/2015-06/2019
B.S. in Information and Computing Science, School of Mathematical Sciences; University of Electronic Science and Technology of China (UESTC)
06

Honors and Awards

  • 2nd Place, ECCV MIPI Challenge on Under-Display Camera (UDC) Image Restoration, 2022
  • National Graduate Scholarship, Ministry of Education, China, 2021
  • National First Prize, Contemporary Undergraduate Mathematical Contest in Modeling (CUMCM), 2017

Academic Activities

Journal & Conference Reviewer:

CVPRTPAMITIPTCYBTMMTIFSTGRSGRSLTCI
《底层视觉之美》