Datasets¶

torch-pointcloud provides loaders for the common point cloud benchmarks. Each sample is one dict, the format transforms consume.

Datasets with an automatic download take download=True; the others must be downloaded manually. S3DIS, ScanNet and ScanObjectNN are released under terms of use: read them (each class links its terms_url), then pass accept_terms=True next to download=True. The original files are kept in root/<dataset>/raw, and the cache built from them in processed.

from torch_pointcloud.datasets import ModelNet40
from torch_pointcloud.utils.data import PointCloudDataLoader

dataset = ModelNet40(root="data", train=True, download=True)
dataloader = PointCloudDataLoader(dataset, batch_size=32)

PointCloudDataLoader is a DataLoader using the packed-batch collate of torch_pointcloud.utils.data.

Tasks¶

Classification

ModelNet and ScanObjectNN: meshes, presampled clouds, difficulty variants.
Segmentation

Indoor rooms and outdoor LiDAR, their splits, and the label conventions.
Part segmentation

ShapeNetPart: 16 categories, 50 global part ids.
Detection

SUN RGB-D, KITTI, nuScenes, and batching ragged boxes.

Object classification¶

Dataset	Samples	Classes
ModelNet10 / ModelNet40 3D ShapeNets: A Deep Representation for Volumetric Shapes	~12k	10 / 40
ModelNet40Hdf5 PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation	~12k	40
ScanObjectNN Revisiting Point Cloud Classification: A New Benchmark Dataset and Classification Model on Real-World Data	2.9k	15

Indoor scene segmentation¶

Dataset	Scenes	Classes
S3DIS 3D Semantic Parsing of Large-Scale Indoor Spaces	272 rooms, 6 areas	13
ScanNet v2 ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes	1.5k scenes	20 (NYU40) / 200

Outdoor / driving segmentation¶

Dataset	Frames	Classes
SemanticKITTI SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences	43k frames	19
Semantic3D Semantic3D.net: A new Large-scale Point Cloud Classification Benchmark	30 scenes	8
Toronto3D Toronto-3D: A Large-scale Mobile LiDAR Dataset for Semantic Segmentation of Urban Roadways	4 areas	8
ParisLille3D Paris-Lille-3D: a large and high-quality ground truth urban point cloud dataset for automatic segmentation and classification	3 scenes	9
DALES DALES: A Large-scale Aerial LiDAR Data Set for Semantic Segmentation	40 tiles	8

Part segmentation¶

Dataset	Samples	Classes
ShapeNetPart A Scalable Active Framework for Region Annotation in 3D Shape Collections	~16k	16 categories / 50 parts

Base class¶

Dataset	Task
PointCloudDataset Base class of the loaders: the `root/<name>/raw` and `processed` paths and their existence checks. It has no download, process or transform hook; a loader implements `__getitem__` and `__len__`, and any `torch.utils.data.Dataset` returning sample dicts works as well.	(any)

Dict keys¶

Every loader returns sample dicts keyed by DataKeys (torch_pointcloud.utils.data): pos, color, normal, segment, instance, label, box, face, ... The data format page lists each key with its shape and dtype, and how collate batches it; the task pages below note what each dataset adds.