Visual Restoration
Developing restoration models that improve visual quality, robustness, and usability in real-world conditions.
Computer Vision · Image Restoration · Multimodal Perception
Vision for visual restoration and multimodal perception.
I received my Ph.D. degree in Computer Science from Wuhan University in 2024 under the supervision of Prof. Lefei Zhang and Prof. Bo Du. I am currently a Postdoctoral Researcher at MBZUAI, mentored by Prof. Salman Khan.
I was a research intern at Horizon Robotics, working with Dr. Qian Zhang and Dr. Guoli Wang, and a visiting student in LV-Lab at NUS under the supervision of Prof. Xinchao Wang.
📧 I am open to collaboration and welcome inquiries from anyone interested in my research topics.
Feel free to reach out via jiaqi.ma AT mbzuai.ac.ae.
Raw Image Processing
Exploring how visual signals can be processed, enhanced, and represented across computational imaging pipelines.
Remote Sensing
Building visual models for large-scale sensing, environmental understanding, and spatial intelligence.
Multimodal Perception
Connecting visual information with other modalities to support richer and more flexible machine perception.
Updates
News
InterLight was accepted to IJCAI 2026.
Token Expand-Merge was accepted to IEEE RA-L.
Won 4th place in the first real-world all-in-one image restoration challenge by LOVIF at CVPR 2026.
Served as one of the organizers of the 11th NTIRE 2026 Efficient Super-Resolution Challenge.
RainDiff was accepted by CVPR 2026.
Simtoken was accepted by ICASSP 2026.
Token Expand-Merge, ClusIR, and EvoIR were released on arXiv.
Perceive-IR was accepted by TIP.
HyperSIGMA was accepted by TPAMI.
Selected Works
Publications
* Equal contribution, ^ Corresponding authors.
Cited by 89
Perceive-IR: Learning to Perceive Degradation Better for All-in-One Image Restoration
Fine-grained quality control for restoring images to more closely resemble their undistorted counterparts across degradation types and severities.
Cited by 222
HyperSIGMA: Hyperspectral Intelligence Comprehension Foundation Model
A billion-parameter-scale vision transformer foundation model for hyperspectral image interpretation.
Cited by 115
ProRes: Exploring Degradation-aware Visual Prompt for Universal Image Restoration
Introduces degradation-aware visual prompts for universal image restoration tasks.
Cited by 28
Restoration and Enhancement on Low Exposure Raw Images by Joint Demosaicing and Denoising
Joint demosaicing and denoising for restoring and enhancing low exposure raw images.
Cited by 12
ELMformer: Efficient Raw Image Restoration with a Locally Multiplicative Transformer
An efficient transformer framework for raw image restoration.
Recognition
Honors and Awards
- China National Scholarship, Ministry of Education of China.
- CSC Scholarship, Ministry of Education of China.
- Yisheng Scholarship, Wuhan University.
- Outstanding Postgraduate, Wuhan University.
- First-Class Outstanding Scholarship for Postgraduate, Wuhan University.
Background
Education
- Ph.D. in Computer Science, Wuhan University.
- Master's degree in Computer Science, Wuhan University.
- Bachelor's degree in Computer Science, Zhengzhou University.
- Research Intern, Horizon Robotics.