Label generation methods for surface models currently voxelize objs from manually created or automatically created segmentations. These can be relatively accurate, but require the segmentation is already created, creating a bit of a catch-22 in that we need labels from hard-to-segment regions that likely do not contain segmentations.
Label generation methods for fibers and ink are entirely manual.
Input Format: zarr or other on-disk npy array
Output Format: zarr
Result: Image/label pairs of surface or fiber labels that are higher quality than simple voxelization of existing obj meshes
Label generation methods for surface models currently voxelize objs from manually created or automatically created segmentations. These can be relatively accurate, but require the segmentation is already created, creating a bit of a catch-22 in that we need labels from hard-to-segment regions that likely do not contain segmentations.
Label generation methods for fibers and ink are entirely manual.
Input Format: zarr or other on-disk npy array
Output Format: zarr
Result: Image/label pairs of surface or fiber labels that are higher quality than simple voxelization of existing obj meshes