招收对象
- 本科生：南京大学及其他高校的大一、大二、大三学生，需保证至少一年的进组科研时间，特别优秀者提供适当科研补助。
- 研究生：硕士/博士研究生，背景包括计算机、数学等相关专业。
- 研究人员：全职/兼职研究助理。
 |
Kai Zhang (张 凯) |
课题组长期招收优秀的同学加入：
关于招收硕士研究生的说明
- 招收对象是愿意全身心投入做好科研、学习专业知识的学生。
- 可以接受论文精读理解、代码实现、PPT报告等考核。
- 对自己的编程能力有自信，并且有一定的数学功底。
- 如成功保研，愿意在课题组内做大四毕设，提前熟悉相关背景知识，并且在研究生入学之前投稿一篇文章。
关于招收博士研究生的说明
- 以第一作者身份在CCF A类会议或期刊发表至少一篇论文。
- 具有生成式大模型相关研究经验者优先考虑。
Biography
Kai Zhang is an Associate Professor at the School of Intelligence Science and Technology, Nanjing University (Suzhou Campus). He received his Ph.D. degree in Computer Science and Technology from Harbin Institute of Technology in 2019. During his Ph.D. studies, he spent three years as a research assistant at The Hong Kong Polytechnic University. From 2020 to 2024, he was a postdoctoral researcher at the Computer Vision Lab, ETH Zurich. He joined Nanjing University in March 2024. His research interests include low-level vision, with a particular focus on image and video restoration and enhancement. His publications have received over 46,000 citations on Google Scholar, with his most cited paper receiving more than 10,000 citations and nine papers receiving over 1,000 citations each. His representative works include DnCNN, FFDNet, SRMD, IRCNN, USRNet, DPIR, SwinIR, BSRGAN, SCUNet, and ConverseNet.
张凯，南京大学苏州校区智能科学与技术学院副教授，博士生导师。2019年9月在哈尔滨工业大学获得博士学位，期间在香港理工大学交流三年，并于2020年1月至2024年2月在瑞士苏黎世联邦理工大学计算机视觉实验室从事博士后研究。2024年3月加入南京大学，专注于图像视频复原与增强等底层视觉问题的研究，论文谷歌学术引用46,000次，单篇最高引用10,000余次，9篇论文引用超1,000次。代表性工作包括DnCNN、FFDNet、SRMD、IRCNN、USRNet、DPIR、SwinIR、BSRGAN、SCUNet以及ConverseNet等。获得的奖项包括2026年国际基础科学大会前沿科学奖，2024年IEEE信号处理学会最佳论文奖，2024年中国电子学会自然科学一等奖，以及2020年黑龙江省自然科学一等奖等。目前主持包括优秀青年科学基金（海外）在内的多项科研项目。
Research Interest
I work in the field of image processing, with a specific focus on developing deep learning techniques for inverse problems in low-level computer vision. Recently, I focus on the following research topics:- Generative methods for low-level vision problems
- Learned camera ISP pipeline
- Denoising and super-resolution
- Deep plug-and-play/unfolding image restoration
- Algorithm and physics-guided network design
Representative Low-Level Vision Applications
Image Denoising
Super-Resolution
RAW Denoising
White Balance
AI-ISP
Light Enhancement
Color Transfer
PyTorch Toolbox for Image Restoration
- KAIR (support training and testing for DnCNN, FFDNet, SRMD, USRNet, ESRGAN)
- DPIR (Plug-and-Play Image Restoration with Deep Denoiser Prior)
News
- (2026-08) 获2026国际基础科学大会(ICBS)信息科学和工程领域前沿科学奖(Frontiers of Science Award in Information Science and Engineering).
- (2025-04) FFDNet(TIP 2018)获IEEE信号处理协会2024最佳论文奖(Best Paper Award).
- (2025-06) I will serve as a Senior Program Committee (SPC) Member for AAAI 2026.
- (2024-06) I will serve as a Senior Program Committee (SPC) Member for AAAI 2025.
- We are organizing a Special Issue titled "Recent Advances in Image Reconstruction" for Visual Intelligence. Submissions of high-quality articles are warmly welcome.
- Four papers are accepted in CVPR 2024.
- (2022-07) I will serve as a Senior Program Committee (SPC) Member for AAAI 2023.
- I am organizing a Special Issue "Image Denoising and Image Super-Resolution for Sensing Application" on Sensors (IF 3.576), submit your manuscript before 15 December 2022.
- We released the testing codes of SCUNet.
- We released the training codes of BSRNet and SwinIR.
- All the four submitted papers are accepted by ICCV 2021. Congratulations to Jingyun Liang and Jiaxi Jiang.
- One paper is accepted by IEEE TPAMI.
- Two papers are accepted in CVPR 2021.
- The code of BSRNet is available.
- The code of DPIR is available.
- The code of USRNet is available.
- I am co-organizing the ECCV 2020 Workshop on Advances in Image Manipulation
- Two papers are accepted in CVPR 2020.
- I am co-organizing the CVPR 2020 Workshop on New Trends in Image Restoration and Enhancement
Selected Publications
 |
ColorFM: An Optimization-to-Learning Framework for Color Transfer via Flow Matching |
 |
Reverse Convolution and Its Applications to Image Restoration |
 |
MoVideo: Motion-Aware Video Generation with Diffusion Models |
 |
Practical Blind Image Denoising via Swin-Conv-UNet and Data Synthesis |
 |
Recurrent Video Restoration Transformer with Guided Deformable Attention |
 |
Towards Interpretable Video Super-Resolution via Alternating Optimization |
 |
Plug-and-Play Image Restoration with Deep Denoiser Prior |
 |
Deep plug-and-play and deep unfolding methods for image restoration (Book chapter) |
 |
SwinIR: Image Restoration Using Swin Transformer |
 |
Designing a Practical Degradation Model for Deep Blind Image Super-Resolution |
 |
Hierarchical Conditional Flow: A Unified Framework for Image Super-Resolution and Image Rescaling |
 |
Mutual Affine Network for Spatially Variant Kernel Estimation in Blind Image Super-Resolution |
 |
Towards Flexible Blind JPEG Artifacts Removal |
 |
Flow-based Kernel Prior with Application to Blind Super-Resolution |
 |
AIM 2020 Challenge on Efficient Super-Resolution: Methods and Results |
 |
Deep Unfolding Network for Image Super-Resolution |
 |
Neural Blind Deconvolution Using Deep Priors |
 |
NTIRE 2020 Challenge on Perceptual Extreme Super-Resolution: Methods and Results |
 |
AIM 2019 Challenge on Constrained Super-Resolution: Methods and Results |
 |
Deep Plug-and-Play Super-Resolution for Arbitrary Blur Kernels |
 |
Learning a Single Convolutional Super-Resolution Network for Multiple Degradations |
 |
Learning Deep CNN Denoiser Prior for Image Restoration |
 |
FFDNet: Toward a Fast and Flexible Solution for CNN-based Image Denoising |
 |
Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising |
 |
Toward Convolutional Blind Denoising of Real Photographs |
 |
Convolutional Neural Networks for Image Denoising and Restoration (Book chapter) |
 |
Joint Learning of Multiple Regressors for Single Image Super-Resolution |
 |
Revisiting Single Image Super-Resolution Under Internet Environment: Blur Kernels and Reconstruction Algorithms |
