fastdev.datasets.modelnet

Module Contents

class fastdev.datasets.modelnet.ModelNetDatasetConfig

Configuration for ModelNetDataset.

data_root: str

download_if_not_exist: bool = False

preprocess_data: bool = True

recache: bool = False

num_categories: Literal[10, 40] = 40

num_points: int = 1024

resample: bool = True

return_normals: bool = False

__post_init__()

class fastdev.datasets.modelnet.ModelNetDataset(config: ModelNetDatasetConfig, split: Literal['train', 'test'] = 'train')

Bases: torch.utils.data.Dataset

ModelNet dataset.

• By setting download_if_not_exist=True, the dataset will be downloaded from Hugging Face Hub.

• The modelnet40 dataset has already been FPS sampled, no need to run FPS sampling again.

Parameters:

• config (ModelNetDatasetConfig)

• split (Literal['train', 'test'])

config

__len__()

__getitem__(index)

static download_data(data_root: str)

Parameters:

data_root (str)