Wu Chi Hsuan (Jason)

I am a PhD student in Computer Science at UT Austin, advised by Prof. Kristen Grauman. I received my bachelor degree from Hong Kong University of Science and Technology in 2023. I worked on my bachelor thesis in the Vision and System Design Lab (VSDL) supervised by Prof. Kwang-Ting Cheng. I also worked as a Research Assistant in the Information and Network Dynamics Lab at EPFL supervised by Prof. Matthias Grossglauser during exchange.

Email  /  CV  /  Medium  /  Github

profile photo

Research

I am interested in Video Understanding and Behavioral Analysis with multi-modalities. While tons of skill-related videos are available online, how to effectively utilize them to assess human skills, generating support or feedback from AI agents, and support skill learning remains underexplored. My current research focuses on bridging knowledge from skilled instructional videos to skill learning applications.

skillsight SportSkills: Physical Skill Learning from Sports Instructional Videos

Kumar Ashutosh, Wu Chi hsuan, Kristen Grauman
arXiv 2026
project page
skillsight SkillSight: Efficient First-Person Skill Assessment with Gaze

Wu Chi hsuan, Kumar Ashutosh, Kristen Grauman
CVPR 2026
project page
stitch-a-demo Stitch-a-Demo: Video Demonstrations from Multistep Descriptions

Wu Chi hsuan*, Kumar Ashutosh*, Kristen Grauman
CVPR 2026
project page
mocorank CMOSE: Comprehensive Multi-Modality Online Student Engagement Dataset with High-Quality Labels

Wu Chi hsuan, Liu Shih Yang, Huang Xijie, Prof. Tim Kwang-Ting Cheng
CVPRW 2024
project page
biasometer It's All Relative: Interpretable Models for Scoring Bias in Documents

Aswin Suresh, Wu Chi hsuan, Prof. Matthias Grossglauser
EACL 2024

Other Projects
CycleGAN Day-Night-Transfomation-for-improving-feature-matching
Computer Vision Project
[Github]

We transformed illumination of day-night image pairs using CycleGAN to improve feature matching.
SR Super-Resolution on Computer Texts
Undergraduate Research Opportunity Program, Supervised by Prof. Qifeng Chen
[Github]

We designed two-stream model to simultaneuously improve text boundary clarity and colors on text images.

Miscellanea

Blog Posts

 Series of Life Recording: Hong Kong Universiy of Science and Technology

Thanks Jon Barron for providing the source code