config.yaml file is missing from the step python fusion.py

[NagaPadma]

[king1111sadjfoisja]

same problem 应在在fusion.py也要输出一下yaml文件

[sharinka0715]

Hi, I did not get the problem. Could you please describe your problem more specifically? Thanks.

[NagaPadma]

Hi, I did not get the problem. Could you please describe your problem more specifically? Thanks.

when i run python distill.py its looking for this yaml in fusion output folder.

PS: i have trained my model on Gaussian splatting original garden scene.

[sharinka0715]

Hi, I did not get the problem. Could you please describe your problem more specifically? Thanks.

when i run python distill.py its looking for this yaml in fusion output folder.

PS: i have trained my model on Gaussian splatting original garden scene.

Code in distill.py does not search for a config.yaml. The config.yaml in your code comes from the wrong parameters of config.scene.scene_dir.

distill.py can never run on a single scene, instead, it runs on a large-scale dataset such as ScanNet.

If you just want to try our semantic gaussians on a single scene, you do not need to conduct 3D distillation. You can just run view_viser.py to view performances on 2D projection results.

[Sugar55888]

@sharinka0715 Hi, so we need to run train.py and fusion.py for each scene in Scannet dataset. And then we can run distill.py to get the output 3D semantic network checkpoints. Is that right?

[sharinka0715]

@sharinka0715 Hi, so we need to run train.py and fusion.py for each scene in Scannet dataset. And then we can run distill.py to get the output 3D semantic network checkpoints. Is that right?

@Sugar55888 Yes, that's right.

[Sugar55888]

@sharinka0715 Thank you!