Yushan Zhang

I'm a WASP funded PhD Student at Linköping Universion in Sweden, supervised by Prof. Michael Felsberg. I was a research intern at NVIDIA Dynamic Vision and Learning Group lead by Laura Leal-Taixe.

I'm interested in 3D Computer Vision in general, including 3D point cloud perception, Scene flow estimation, Computer vision for robotics, 3D scene understanding, Open-vocabulary object segmentation and tracking, Diffusion models, Vision-language models etc.

Email  /  CV  /  Scholar  /  Github

profile photo

Research

fast-texture Zero-shot 4D Lidar Panoptic Segmentation
Yushan Zhang, Aljoša Ošep, Laura Leal-Taixé, Tim Meinhardt
CVPR, 2025
paper / code / project page

We propose SAL-4D (Segment Anything in Lidar--4D), a method that utilizes multi-modal sensory robotic setups as a bridge to distill recent developments in Video Object Segmentation (VOS) in conjunction with off-the-shelf Vision-Language foundation models to Lidar.

fast-texture DiffSF: Diffusion Models for Scene Flow Estimation
Yushan Zhang, Bastian Wandt, Maria Magnusson, Michael Felsberg
NeurIPS, 2024 (Spotlight)
paper / code / project page

Introduce diffusion models for scene flow estimation, enabling uncertainty estimation.

fast-texture GMSF: Global Matching Scene Flow
Yushan Zhang, Johan Edstedt, Bastian Wandt, Per-Erik Forssén, Maria Magnusson, Michael Felsberg
NeurIPS, 2023
paper / code / project page

Single-scale one-shot global matching for scene flow estimation.