My research is related to 3D scene understanding, camera calibration and robotics.
I was previously a Master student in Robotics. Before that, I received my B.S.E. degrees in Computer Science at UM and Mechanical Engineering at Shanghai Jiao Tong University through a dual degree program at UM-SJTU Joint Institute.
- [2025/03] We've released Stereo4D dataset and MegaSaM code.
- [2025/02] Stereo4D and MegaSaM are accepted to CVPR 2025!
- [2024/02] Two papers accepted to CVPR 2024!
- [2024/02] I will join Google as a Student Researcher, working with Noah Snavely and Aleksander Hołyński.
Our new system for scene-level 3D reconstruction from posed images, which works with as few as one view, reconstructs the complete geometry of unseen scenes, including hidden surfaces.
Learning 3D implicit function from a single input image. Unlike other methods, D2-DRDF does not depend on mesh supervision during training and can directly operate with raw RGB-D data obtained from scene captures.
We introduce a simpler approach that uses a transformer applied to 3D-aware plane tokens to perform 3D reasoning. This is substantially more effective than SparsePlanes.
We learn to reconstruct scenes from sparse views with an unknown relationship. We take advantage of planar regions and their geometric properties to recover the scene layout.
We augment a manipulation planner for cluttered environments with a shape completion network and a volumetric memory system, allowing the robot to reason about what may be contained in occluded areas.