torch_geometric.datasets.ShapeNet

class ShapeNet(root: str, categories: Optional[Union[str, List[str]]] = None, include_normals: bool = True, split: str = 'trainval', transform: Optional[Callable] = None, pre_transform: Optional[Callable] = None, pre_filter: Optional[Callable] = None, force_reload: bool = False)[source]

Bases: InMemoryDataset

来自“A Scalable Active Framework for Region Annotation in 3D Shape Collections”论文的ShapeNet部件级别分割数据集，包含来自16个形状类别的约17,000个3D形状点云。每个类别都标注了2到6个部件。

Parameters:

  • root (str) – Root directory where the dataset should be saved.

  • categories (str 或 [str], 可选) – CAD模型的类别 （可以是 "Airplane", "Bag", "Cap", "Car", "Chair", "Earphone", "Guitar", "Knife", "Lamp", "Laptop", "Motorbike", "Mug", "Pistol", "Rocket", "Skateboard", "Table" 中的一个或多个组合）。 可以显式设置为 None 以加载所有类别。 （默认值：None）

  • include_normals (bool, optional) – 如果设置为 False，将不会 将法向量作为输入特征包含在 data.x 中。 因此，data.x 将为 None。 （默认值：True）

  • split (str, optional) – 如果为 "train"，则加载训练数据集。 如果为 "val"，则加载验证数据集。 如果为 "trainval"，则加载训练和验证数据集。 如果为 "test"，则加载测试数据集。 (默认值: "trainval")

  • transform (callable, optional) – A function/transform that takes in an torch_geometric.data.Data object and returns a transformed version. The data object will be transformed before every access. (default: None)

  • pre_transform (callable, optional) – A function/transform that takes in an torch_geometric.data.Data object and returns a transformed version. The data object will be transformed before being saved to disk. (default: None)

  • pre_filter (callable, optional) – A function that takes in an torch_geometric.data.Data object and returns a boolean value, indicating whether the data object should be included in the final dataset. (default: None)

  • force_reload (bool, optional) – Whether to re-process the dataset. (default: False)

统计:

#图表

#节点

#edges

#特性

#classes

16,881

~2,616.2

0

3

50