Biography
I am a final-year PhD student at the State University of New York at Buffalo, where I am advised by Prof. David Doermann and Prof. Junsong Yuan. Prior to this, I received my B.Eng in Computer Science from Xi’an Jiaotong University (XJTU) in 2020. During my undergraduate study at XJTU, I was fortunate to be advised by Ziyi Liu and Prof. Le Wang at the Institute of Artificial Intelligence and Robotics.
The best way to reach me is through my email: yuanhaozhai [AT] gmail [DOT] com. My Chinese name is 翟元浩 (pronunciation).
Here is my research proposal on learning human action.
🚨I am on the job market and looking for a US-based Research/Applied Scientist position starting from summer 2025. Feel free to contact me if you have any openings!
- Video Understanding/Synthesis
- Human Action Analysis/Synthesis
- Multi-Modal Learning
- Weakly-Supervised Learning
- Media Forensics
-
PhD in Computer Vision, 2021 - Present
State University of New York at Buffalo, Buffalo, USA
-
B.Eng in Computer Science, 2016 - 2020
Xi'an Jiaotong University, Xi'an, China
-
Exchange Student, 2019
University of California, Berkeley, Berkeley, USA
News
- 09/2024: Motion consistency model was accepted to NeurIPS 2024!
- 07/2024: 💃 One paper on human-centric joint video-depth generation is accepted to ECCV 2024. Check out our project page here!
- 06/2024: 🚀 We present motion consistency model, accelerating video diffusion while improving frame quality. Check out our project page, paper, and demo!
- 05/2024: One paper on neural architecture search is accepted to ICML 2024; congrats to all collaborators! Check out our paper here and code here!
- 04/2024: We propose STAT, a pioneering method for the generalizable temporal action localization task. Read our paper here!
- 03/2024: Release the Gradio demo of our image manipulation detection model.
- 02/2024: Our human dancing video generation paper DisCo is accepted to CVPR 2024, congratulations to the collaborators!
- 07/2023: One paper on human motion synthesis is accepted to ACM MM 2023.
- 07/2023: Two papers are accepted to ICCV 2023. One is on open-set action recognition, and the other is on image manipulation detection.
- 02/2023: One paper on 3D hand shape reconstruction is accepted to CVPR 2023.
- 07/2022: One paper on weakly-supervised temporal action localization is accepted to TPAMI.
- 06/2021: We won the 2nd place in the HACS W-TAL Challenge 2021; check out our report.
Experience
