Visualization of STPLS3D semantic segmentation

[whuhxb]

Hi @meidachen

If using block generation upon STPLS3D dataset, how to obtain the visualization results of the original whole area of STPLS3D dataset? Merge? Or?

Thanks a lot.

[RockyatASU]

Thanks for your interest in our datasets. As the outputs of the point transformer are only labels, you firstly need to merge predicted labels into the input point clouds, i.e., the generated blocks and then directly merge all the blocks to generate the segmentation results of the original whole area. All the mentioned merging operations above can be done by numpy functions like concatenation and using the write_ply function under the util folder to save the whole are point clouds with labels. We will upload another script to merge the generated blocks and store the point cloud with labels in .ply files. You can visualize the point cloud in some point cloud viewers like CloudCompare.

[whuhxb]

@RockyatASU
OK Thanks a lot.

[RockyatASU]

@whuhxb Mentioned script has been uploaded.