kaolin.io.shrec

API

class kaolin.io.shrec.SHREC16(root, categories=None, split='train', transform=None, output_dict=False)

Bases: Dataset

Dataset class for SHREC16, used for the “Large-scale 3D shape retrieval from ShapeNet Core55” contest at Eurographics 2016. More details about the challenge and the dataset are available here.

The __getitem__ method will return:

  • if output_dict=True: a dictionary with the following key-value pairs:

    • ‘mesh’: containing a kaolin.rep.SurfaceMesh returned by kaolin.io.obj.import_mesh().

    • ‘name’: the model name (i.e the subfolder name)

    • ‘path’: the full path to the .off

    • ‘synset’: the synset associated to the category

    • ‘labels’: the labels associated to the category (see synset_to_labels)

  • if output_dict=False (deprecated): a KaolinDatasetItem with the fields:

    • data: containing a kaolin.rep.SurfaceMesh returned by kaolin.io.obj.import_mesh().

    • attributes: containing a dictionary with the following key-value pairs:

      • ‘name’: the model name (i.e the subfolder name)

      • ‘path’: the full path to the .off

      • ‘synset’: the synset associated to the category

      • ‘labels’: the labels associated to the category (see synset_to_labels)

Deprecated since version 0.13.0: output_dict=False is deprecated. Datasets should always output a dictionary to be compatible with ProcessedDataset.

Parameters

  • root (str) – Path to the root directory of the dataset.

  • categories (list) – List of categories to load (each class is specified as a string, and must be a valid SHREC16 category). If this argument is not specified, all categories are loaded by default.

  • split (str) – String to indicate what split to load, among [“train”, “val”, “test”]. Default: “train”.

  • transform (Callable) – A function/transform that takes in a dictionary or KaolinDatasetItem and returns a transformed version.

  • output_dict (bool) – If True, __getitem__ output a dictionary, else KaolinDatasetItem (deprecated) Default: False.

get_attributes(index)
get_cache_key(index)
get_data(index)