Wanzhou Liu / 刘万洲

CS PhD student at Pitt.

prof_pic.jpg

"I will give it back when I get there."

Welcome to my website! I am currently a first-year CS PhD student at Pitt, grateful to be advised by Prof. Adriana Kovashka. Previously, I was a MS student at WashU, where I luckily worked with Prof. Nathan Jacobs. I was also fortunate to collaborate with Prof. Eshed Ohn-Bar at Boston University. I received my BEng from Qingdao University.

I am intrigued by cutting-edge computer vision techniques, including multimodal large language models, 3D spatiotemporal perception, computational photography, and generative models. I pursue both theoretical and practical innovations, valuing simplicity, efficiency, and scalability in impactful research.

Deeply inspired by Prof. Jeremy Howard, I have an ultimate dream to establish a nonprofit organization akin to Fast.ai, which actually launched my journey into machine learning. I believe this begins with a commitment to helping others and fostering collaboration. If you have any questions or ideas in mind, please feel free to reach out!

Aside from research, I find great joy in literature and photography. You can find some of my photos tucked away in a quiet corner of this web. :)

News

Aug 25, 2025 Formally joined University of Pittsburgh as a PhD student!
Jun 25, 2025 Our DriveQA was accepted to ICCV 2025!

Selected Publications

Symbol * or † represents equal contribution or advising.
  1. ICCV
    passingdriving.png
    DriveQA: Passing the Driving Knowledge Test
    Maolin Wei*Wanzhou Liu*, and Eshed Ohn-Bar
    In IEEE/CVF International Conference on Computer Vision (ICCV) , 2025
  2. CVPRW
    declutternerf.png
    DeclutterNeRF: Generative-Free 3D Scene Recovery for Occlusion Removal
    Wanzhou Liu, Zhexiao Xiong, Xinyu Li, and Nathan Jacobs
    In IEEE/CVF CVPR 2025 CV4Metaverse Workshop (CVPRW) , 2025
  3. ICCAID
    cyclegan.png
    As Time Goes By: Dynamic Face Aging Method Using CycleGAN
    Wanzhou Liu
    In International Conference on Computer Graphics, Artificial Intelligence, and Data Processing (ICCAID) , 2022