Xingyu Zhou

I am currently in my first year of pursuing a doctoral degree at University of Electronic Science and Technology of China (UESTC CVL-Lab). My advisor is Shuhang Gu. I received the B.S. degree from the Artificial Intelligence School, Xidian University in 2023.

My previous research interest is low-level vision, such as image/video restoration, image enhancement and so on. My current research interest lies in image or video generation. However, what remains unchanged is that I have always been excited about how to build more efficient models, including training and inference. In addition, I believe that this will ultimately impact the development of the AI community.

Email  /  Scholar  /  Twitter  /  Github

News

[2026-01] One paper TVQ&RAP is accepted to ICLR 2026.

[2025-06] One paper CTMSR is accepted to ICCV 2025.

[2025-02] Two papers PFT-SR and DCAE are accepted to CVPR 2025.

[2024-03] Three papers MIA-VSR, ATD-SR and FR-INR are accepted to CVPR 2024.

Selected Publications

We achieve a new state of the art on ImageNet. IG delivers dramatic quality gains with far fewer training epochs, adds negligible overhead, and works as a drop-in upgrade for modern diffusion transformers.

Investigate the ability of video super-resolution models to utilize long range temproal information. While enjoying the training efficiency of short video sequences, it achieves results equivalent to training on long video sequences.

Explore temporal redundancy in video super-resolution models. Achieve a significant reduction in the inference computational cost of video super-resolution models while not compromising restoration performance.

Thank you to Jon Barron for the source code for the website!