Leo / Zeqing Yuan
Leo / Zeqing Yuan
Grad Student @MIT

I am pursuing my Ph.D. at MIT. I am interested in vision, robotics and agents. Previously, my master's research focused on physics-based vision and imaging, and my undergraduate work explored generative models and video understanding.

I obtained my M.S. in Robotics from Carnegie Mellon University, where I was fortunate to be advised by Prof. Srinivasa Narasimhan and Prof. Aswin Sankaranarayanan. I also worked with MD Anderson and Rice on AI for precision tumor surgery. Prior to that, I earned my bachelor's degree with the highest honor from Zhejiang University, working with Professors Yingcai Wu, Jake Zhao and Chunhua Shen, and spent a summer interning at Stanford with Prof. Jiajun Wu's group.

× Enlarged view

BibTeX Citation

×

            
          

Publications and Manuscripts

VT-Intrinsic: Physics-Based Decomposition of Reflectance and Shading using a Single Visible-Thermal Image Pair
TLDR Light absorbed rather than reflected leaves thermal traces that reveal lighting and material reflectance.
Revealing Subtle Heat Flows all around us using Microbolometric Videos
TPAMI & ICCP 2026
TLDR Revealing subtle signals in thermal videos by modeling bolometric sensor physics.
SinLayout: Learning Generative 3D Scene Layouts from a Single Image
AI3DG @CVPR 2024
TLDR Learning spatial layout distribution from a single image using an adversarial generator.
FreeCompose: Generic Zero-Shot Image Composition with Diffusion Prior
TLDR Exploiting pretrained diffusion priors for training-free image composition.
ViSTec: Video Modeling for Sports Technique Recognition and Tactical Analysis
TLDR Parsing broadcast racket-sport videos into stroke-level technique sequences.
3D-PreMise: Can Large Language Models Generate 3D Shapes with Sharp Features and Parametric Control?
Tech Report 2023
TLDR A framework and dataset for LLMs to generate code for parametric 3D modeling.
TIVEE: Visual Exploration and Explanation of Badminton Tactics in Immersive Visualizations
TLDR Immersive VR visualization system for badminton tactical analysis.

Talks

[2026/01] Invited talk @Columbia University: Physics-based Vision with Heat and Light

Selected Honors

Best Poster Award, CVPR Workshop on Computational Cameras and Displays, 2025
AAAI Student Scholarship, 2024
Chinese National Scholarship, Ministry of Education, 2021, 2022, 2023
Chu Kochen Presidential Scholarship, Zhejiang University, 2023
Top 10 Outstanding Students Award at Zhejiang University, 2023
SenseTime Scholarship, 2022

Academic Services

Program Committee / Reviewer for AAAI, CVPR, ECCV, NeurIPS
Teaching Assistant for CS 21191070: Computer Vision, 2023
Mentor at CMU Paths to AI Research (PAIR) Program, 2025
Member of CMU School of Computer Science Dean's Advisory Committee, 2025
Conference Volunteer for AAAI, 2024

Miscellaneous

I love recording and understanding how nature paints with light.
I also enjoy a wide range of sports and spent five lovely years on a professional badminton team. During undergrad, I worked on AI for racket sports, which was later adopted by the national team. Here's a snapshot of my athletic badminton career:
Badminton achievements
  • U.S. National Collegiate Team Championships (D1-B) 🥇
  • Chinese National First-Class Athlete
  • Chinese National 8th Place in Men's Singles
  • Zhejiang Provincial 🥇🥇🥈
  • Fujian Provincial 🥈🥈🥉🥉


Top