I am a PhD student in Computer Science at University of Southern California, advised by Prof. Jernej Barbic. I’ve finished my Master degree in CS@USC advised by Prof. Yue Wang and Prof. Jernej Barbic. I have also interned at Vision & Graphics Lab, where I was fortunate to work with Prof. Yajie Zhao.
Interests
- Differentiable Simulation
- Real-to-Sim
- High Fidelity Digital Twin and Simulation
- Robotics, Computer Graphics, 3D Computer Vision
Education
PhD in Computer Science, 2025-
University of Southern California
M.S. in Computer Science, 2023-2025
University of Southern California
News
- [2025-10] One paper is accepted to NeurIPS 2025!
- [2025-07] One paper is accepted to ICCV 2025!
- [2024-12] One paper is accepted as Oral Presentation to 3DV 2025!
- [2024-09] One paper is accepted to WACV 2025!
Publications
.js-id-selected
Seeing the Wind from a Falling Leaf
NeurIPS 2025
Differentiable pipeline that recovers wind forces from leaf videos and enables physics-consistent video editing and wind manipulation.
Learning an Implicit Physical Model for Image-based Fluid Simulation
ICCV 2025
Physics-informed model that turns one fluid image into a 4D, multi-view animation using learned implicit motion fields and Navier–Stokes priors.
DREAM: Differentiable Real-to-Sim-to-Real Engine for Learning Robotic Manipulation
3rd RSS Workshop on Dexterous Manipulation: Learning and Control with Diverse Data
Differentiable real-to-sim-to-real pipeline that calibrates simulation from real data and transfers manipulation policies back to robots.
Geometry-aware Feature Matching for Large-Scale Structure from Motion
3DV 2025
(Oral Presentation)
A novel geometry-aware feature matching approach that combines detector-based and detector-free methods to improve correspondence accuracy and density for large-scale Structure from Motion.
Volume Rendering of Human Hand Anatomy
arXiv
We propose novel transfer functions for volumetric rendering of hand MRI data that improve visualization of complex hand anatomy while maintaining fine control over tissue appearance.
Skyeyes: Ground Roaming using Aerial View Images
WACV 2025
Skyeyes is a framework that can generate photorealistic sequences of ground view images using only aerial view inputs, thereby creating a ground roaming experience.
Understanding street-level urban vibrancy via spatial-temporal Wi-Fi data analytics: Case LivingLine Shanghai
Environment and Planning B: Urban Analytics and City Science
Using Wi-Fi data analytics and machine learning to quantify urban vibrancy at the street level, demonstrated through a case study of urban interventions in Shanghai while preserving user privacy.
Refined self-attention mechanism based real-time structural response prediction method under seismic action
Engineering Applications of Artificial Intelligence
Our SeisFormer model based on self-attention mechanism for real-time structural response prediction under seismic action for large-scale structures, achieving high accuracy and efficiency even with limited training data.
Real‐time automatic crack detection method based on drone
Computer-Aided Civil and Infrastructure Engineering
A real-time drone-based crack detection method that combines lightweight and high-precision algorithms with autonomous flight control for efficient building damage assessment.
A three-stage deep-learning-based method for crack detection of high-resolution steel box girder image
Smart Structures and Systems
(Excellent Award@IPC-SHM'2020)
A novel three-stage deep learning method for crack detection in high-resolution steel box girder images, combining patch-based CBAM ResNet-50 for localization, Attention U-Net for edge detection, and morphological operations for refinement.
Experience
Research Intern
Research on differentiable simulation for scene reconstruction and video generation; resulted in a NeurIPS’25 paper and an ICCV’25 paper.
Research Intern
Research on urban scence reconstruction and generation based on aerial imagery, which resulted in a WACV'25 paper and a 3DV'25 paper.
Worked on computer vision for structural health monitoring; led to papers in Eng. Appl. AI, CA-CIE, and SSS.
Research Intern
Developed pedestrian trajectory analysis and mined patterns; led to the Environment and Planning B paper on street-level urban vibrancy.
Fun Facts
My Erdős number is 3: Zhiyuan Gao → Jitendra Malik → Fan Chung Graham → Paul Erdős.
I built phdstat, a tool for visualizing grad school application data from TheGradCafe with near real-time updates. It has helped 1k+ applicants track their applications. If you’re interested in contributing to this project, feel free to contact me.
I've contributed to several games (click to expand)
Flappy BalloonAlt-controller party game guiding a balloon through obstacles with custom inputs.I was one of the producers.Trailer: https://vimeo.com/1139208458
PL-23AI-driven detective/puzzle game in the spirit of story-rich investigation titles.I am the lead producer and a past backend engineer.Trailer: coming soon.
Contact
- gaozhiyu@usc.edu
- 300 N Beaudry Ave, Los Angeles, CA 90012