Meshes

Like skeletons, meshes are also computed on SegmentationObject level and can be merged for SuperSegmentationObjects, which agglomerate SegmentationObject instances.

Mesh computation is implemented in syconn.proc.meshes and based open3d and zmesh.

SSO meshes can be called like sso.load_mesh(obj_type) whereobj_type can be in ['sv', 'mi', 'sj', 'vc'] or directly from properties sso.mesh, sso.mi_mesh, … For SO you can access meshes via so.mesh, if they do not exist yet they will be computed and cached automatically.

For caching all object meshes and of SSVs in a SuperSegmentationDataset (SSD) see mesh_preprocessing.py in SyConn/scripts/backend. This is run on a single node with 20 cpus by default (sufficiently fast).