acronym icon indicating copy to clipboard operation
acronym copied to clipboard

Published 20 hours ago • verified publisher icon NVlabs

Parsing of ShapeNetSem meshes

Open gchal opened this issue 4 years ago • 6 comments
trafficstars

Can you give some better instructions regarding the mesh creation with manifold? I am following your steps and leads to segmentation fail. Is the shapenetsem file models-OBJ.zip the one needed or am I doing sth wrong?

gchal avatar Dec 04 '20 15:12 gchal

Hi,

  1. What exactly leads to a segfault? Is it the manifold or simplify command? (The simplify command can lead to crashes when dealing with degenerated triangles, see e.g.: https://github.com/hjwdzh/Manifold/issues/5)
  2. Does it happen with all meshes or just a subset?

clemense avatar Feb 25 '21 16:02 clemense

@clemense I am dealing with problem #1. How did you deal with it?

seann999 avatar Apr 27 '21 00:04 seann999

The easiest solution is to ignore those meshes. There shouldn't be many that fail.

clemense avatar Apr 27 '21 01:04 clemense

Got it; with some patience, I got around 5250 successfully simplified meshes (some seemed to take a really long time, which I killed and discarded)

seann999 avatar Apr 27 '21 05:04 seann999

@clemense I have the same error and am getting segmentation faults for objects in the ShapeNetSem obj files, which should be the same objects that are in acronym. for example, I am getting a crash for 5bbcf08fabf2d50336b41ce3eef1f5be.obj, a "DoubleBed" object. In this case do you suggest (1) removing such objects from the dataset, (2) using the non-simplified version (i.e. the version that was made watertight by the manifold command but not simplified), or (3) something else? Thank you for your help.

thomasweng15 avatar Dec 01 '21 17:12 thomasweng15

It seems that (1) would require regenerating scene_contacts as some scenes reference the removed meshes.

thomasweng15 avatar Dec 01 '21 18:12 thomasweng15