minghanqin / LangSplat

Official implementation of the paper "LangSplat: 3D Language Gaussian Splatting" [CVPR2024 Highlight]
https://langsplat.github.io/
Other
674 stars 72 forks source link

Simultaneous training without checkpoint #61

Closed alicaank closed 3 months ago

alicaank commented 3 months ago

Hi all, I am trying to integrate language features to https://github.com/spla-tam/SplaTAM method, which requires train everything simultaneously. I am working on the Replica dataset and there are my results. As you can see everything getting be same. Do you have any idea or suggestions? Thanks for the great work and help.

Screenshot 2024-07-27 211509 Screenshot 2024-07-27 212337 Screenshot 2024-07-27 213139 Screenshot 2024-07-27 214052