Guofeng Zhang

WildSplat achieves state-of-the-art novel view synthesis from unposed images by effectively decoupling geometry and appearance, even under challenging lighting conditions.

Relationship-aware 3D scene understanding can significantly enhance open-vocabulary segmentation, outperforming traditional methods that ignore object relationships.

OpenReLoc achieves unprecedented relocalization accuracy by leveraging open-vocabulary semantics and structured object representations, setting a new standard for indoor navigation systems.

Disentangling geometry and texture in neural mesh representations unlocks versatile editing capabilities that traditional methods can't match.