What is the meaning of these parameters？loop 、voxel

[swzaaaaaaa]

voxel_size= 0.04,voxel_max = 80000,and loop = 30;What is the meaning of these parameters？

[QuanticDisaster]

I believe the model trains on the voxellized point cloud to reduce density, so voxel_size correspond to the size of voxels, voxel_max the maximum number of voxels/points in a sample (ie a room). For loop, it seems the dataset is "duplicated" or looped through a certain number of times, so the loop would correspond to the number of times it is duplicated I think

[swzaaaaaaa]

Does voxel size mean that there's only one point in a cube, and then there's multiple voxels in a room, and then there's multiple cycles in a room?

[QuanticDisaster]

Yes, this is how I personally understand it

[swzaaaaaaa]

ok，thanks.

[swzaaaaaaa]

Hello, may I ask which paper you refer to for data processing? Is this different from PAConv Repository?

[QuanticDisaster]

I am not the author nor even one of the contributor so I can't tell you for sure but I think they use the same preprocessed data as for PAConv repository (although the dataset class construction is different). For the preprocessing part, in PAConv they seem to use the one from pointnet : CVMI-Lab/PAConv#10

[swzaaaaaaa]

But isn't it different to have every block in PointNet as input, whereas in your replay code, you take voxel subsamples and take the entire room as input?

[aoligei178]

What does step_epoch=30 means in your config file?