Miaowei Wang

He is currently a PhD student in the School of Informatics at the University of Edinburgh, where his research focuses on controllable motion representation in 4D vision and computer graphics. He has strong academic and industrial experience in computer vision, computer graphics and machine learning. His PhD journey is fortunately advised by Prof. Amir Vaxman and Prof. Oisin Mac Aodha.

Previously, he completed his graduate studies in the Department of Electrical Engineering and Computer Science at the University of Michigan, under the supervision of Prof. Jason Corso. He also collaborated closely with Prof. Daniel Morris at Michigan State University on research in 3D vision.

You can find his CV here: 📄 Miaowei’s Curriculum Vitae

📝 Representative Publications (First Authorship)

1. CanFields: Consolidating Diffeomorphic Flows for Non-Rigid 4D Interpolation from Arbitrary-Length Sequences

   Accepted to ICCV 2025

   

   We reconstruct a continuous spatiotemporal manifold from sparse point clouds, addressing the challenge of non-rigid 4D shape interpolation.

   🔗 ArXiv · Project Page · Code

2. DecoupledGaussian: Object-Scene Decoupling for Physics-Based Interaction

   CVPR 2025

   

   A framework for simulating dynamic interactions involving contact and separation, by decoupling object and scene in a physically-aware manner.

   🔗 ArXiv · Project Page · Code

3. Self‑Annotated 3D Geometric Learning for Smeared Points Removal

   WACV 2024

   

   Proposes a self‑supervised method to remove smeared or noisy points in LiDAR/depth‑sensor data by learning consistent 3D geometric structures without manual annotations.

   🔗ArXiv · Code

🎓 Teaching Experiences

Tutor — CGGS: Computer Graphics — Geometry and Simulation

University of Edinburgh Jan – May 2024, Jan – May 2025