Zihui (Sherry) Xue


Hi, I am Zihui Xue (薛子慧), a third-year Ph.D. student at UT Austin, advised by Prof. Kristen Grauman. I am also a visiting researcher at FAIR, Meta AI. Previously, I'm fortunate to work with Prof. Radu Marculescu on efficient deep learning and Prof. Hang Zhao on multimodal learning. I obtained my bachelor's degree from Fudan University in 2020.

My research interests lie in video understanding and multimodal learning.

Email  |  CV  |  Google Scholar  |  Github

profile photo
News
  • [Feb. 2024] Three papers (one first-author) got accepted by CVPR'24. See you in Seattle ☕️.
  • [Sep. 2023] AE2 got accepted by NeurIPS'23. See you in New Orleans 🦪.
  • [Feb. 2023] EgoT2 got accepted by CVPR'23 as Highlight. See you in Vancouver 🏔️.
  • [Jan. 2023] MFH got accepted by ICLR'23 (top-5%).
  • [Aug. 2022] Spent a wonderful summer interning at FAIR, Meta AI, working with Lorenzo Torresani 😊
  • [Sep. 2021] One paper got accepted by NeurIPS'21.
  • [Sep. 2021] One paper got accepted by CoRL'21.
  • [Jul. 2021] Two papers got accepted by ICCV'21.
  • [Aug. 2020] Start working with Prof. Hang Zhao at Shanghai Qi Zhi Institue, Tsinghua University on multimodal learning 😊
Projects
Video Understanding
Learning Object State Changes in Videos: An Open-World Perspective

Zihui Xue, Kumar Ashutosh, Kristen Grauman
CVPR, 2024 [paper] [webpage]
Localization of object state change from videos in the open world
Ego-Exo4D: Understanding Skilled Human Activity from First-and Third-Person Perspectives

Kristen Grauman, Andrew Westbury, Lorenzo Torresani, Kris Kitani, Jitendra Malik, ..., Zihui Xue, et al.
CVPR, 2024 [paper] [webpage] [blog]
A diverse, large-scale multimodal multiview video dataset and benchmark challenge
Detours for Navigating Instructional Videos

Kumar Ashutosh, Zihui Xue, Tushar Nagarajan, Kristen Grauman
CVPR, 2024 [paper]
The video detours problem for navigating instructional videos
Learning Fine-grained View-Invariant Representations from Unpaired Ego-Exo Videos via Temporal Alignment

Zihui Xue, Kristen Grauman
NeurIPS, 2023 [paper] [webpage]
Fine-grained ego-exo view-invariant features -> temporally align two videos from diverse viewpoints
Egocentric Video Task Translation

Zihui Xue, Yale Song, Kristen Grauman, Lorenzo Torresani
CVPR Highlight, 2023 (top-2.5%) [paper] [webpage]
Hollistic egocentric perception for a set of diverse video tasks
Multimodal Learning and Self-supervised Learning
The Modality Focusing Hypothesis: Towards Understanding Crossmodal Knowledge Distillation

Zihui Xue*, Zhengqi Gao* Sucheng Ren*, Hang Zhao
ICLR, 2023 (top-5%) [paper] [webpage]
When is crossmodal knowledge distillation helpful?
Dynamic Multimodal Fusion

Zihui Xue, Radu Marculescu
CVPR MULA workshop, 2023 [paper]
Adaptively fuse multimodal data and generate data-dependent forward paths during inference time.
What Makes Multi-Modal Learning Better than Single (Provably)

Yu Huang, Chenzhuang Du, Zihui Xue, Xuanyao Chen, Hang Zhao, Longbo Huang
NeurIPS, 2021 [paper]
Can multimodal learning provably perform better than unimodal?
Multimodal Knowledge Expansion

Zihui Xue, Sucheng Ren, Zhengqi Gao, Hang Zhao
ICCV, 2021 [paper] [webpage]
A knowledge distillation-based framework to effectively utilize multimodal data without requiring labels.
On Feature Decorrelation in Self-Supervised Learning

Tianyu Hua, Wenxiao Wang, Zihui Xue, Sucheng Ren, Yue Wang, Hang Zhao
ICCV, 2021 (Oral, Acceptance Rate 3.0%) [paper] [webpage]
Reveal the connection between model collapse and feature correlations!
Efficient Deep Learning
SUGAR: Efficient Subgraph-level Training via Resource-aware Graph Partitioning

Zihui Xue, Yuedong Yang, Mengtian Yang, Radu Marculescu
IEEE Transactions on Computers, 2023 [paper]
An efficient GNN training framework that accounts for resource constraints.
Anytime Depth Estimation with Limited Sensing and Computation Capabilities on Mobile Devices

Yuedong Yang, Zihui Xue, Radu Marculescu
CoRL, 2021 [paper]
Anytime Depth Estimation with energy-saving 2D LiDARs and monocular cameras.