Kwang Moo Yi (@kwangmoo_yi)

2025-06-30 | โค๏ธ 160 | ๐Ÿ” 28


Preprint of (not) today: Xu et al., โ€œ4DGT Learning a 4D Gaussian Transformer Using Real-World Monocular Videosโ€ โ€” https://4dgt.github.io/

A feed-forward dynamic Gaussian Splatting estimator trained with monocular videos. 2-stage coarse-to-fine training + mono-depth guidance. https://x.com/kwangmoo_yi/status/1939735405479567759/video/1


See similar notes in domain-vision-3d, domain-llm

Tags

type-paper domain-vision-3d, domain-llm