Could you please provide the code for ScanNet++ and KITTI-360 dataset processing and training?

[Sugar55888]

Thank you!

[samcao0416]

For scanner++, we subsampled mesh to pointclouds. It’s simple with cloud compare/meshlab/open3d…
KITTI360 is more complicate. You need you combine different LiDAR pcds with transform matrices applied. Then we also filter the noises, turn LiDAR pcds into mesh and then downsampled to pointclouds. I remember transforms are also need to be applied on camera poses.

It will take me some days to clean the code. I’m still in Tokyo for attending the conference.

[Sugar55888]

@samcao0416 Thank you very much for your reply! Looking forward to more instructions in the repository.