4D Reconstruction

TimeFormer: Capturing Temporal Relationships of Deformable 3D Gaussians for Robust Reconstruction
We design a plug-and-play module called TimeFormer to enable existing deformable 3D Gaussians reconstruction methods with the ability to implicitly model motion patterns from a learning perspective.
Dadong Jiang, Zhihui Ke, Xiaobo Zhou, Zhi Hou, Xianghui Yang, Wenbo Hu, Tie Qiu, Chunchao Guo
PDF Cite Code Project Slides
TimeFormer: Capturing Temporal Relationships of Deformable 3D Gaussians for Robust Reconstruction