naderalfares/ModelNet40_Auto_aligned

1---2license: mit3task_categories:4- image-classification5tags:6- 3d7- point-cloud8- mesh9- modelnet4010- cad11pretty_name: ModelNet40 Auto Aligned12size_categories:13- 10K<n<100K14configs:15- config_name: default16  data_files:17  - split: train18    path: data/train-*19  - split: test20    path: data/test-*21dataset_info:22  features:23  - name: object_id24    dtype: large_string25  - name: class26    dtype: large_string27  - name: split28    dtype: large_string29  - name: object_path30    dtype: large_string31  - name: __index_level_0__32    dtype: int6433  splits:34  - name: train35    num_bytes: 18643836    num_examples: 984337  - name: test38    num_bytes: 4821539    num_examples: 246840  download_size: 23465341  dataset_size: 23465342---43 44# ModelNet40 Auto Aligned45 46Auto-aligned version of the [ModelNet40](https://modelnet.cs.princeton.edu/) 3D CAD dataset. Each sample is an OFF mesh file organized by class and train/test split.47 48This dataset mirrors the layout of [`naderalfares/ModelNet40`](https://huggingface.co/datasets/naderalfares/ModelNet40), but uses the auto-aligned meshes from the Princeton ModelNet release.49 50## Dataset structure51 52```53modelnet40_auto_aligned/54  {class}/55    train/{class}_{id}.off56    test/{class}_{id}.off57```58 59- **40 classes** (airplane, bathtub, bed, …, xbox)60- **9,843** training meshes61- **2,468** test meshes62- **12,311** meshes total (~9.7 GB)63 64The parquet manifest stores metadata only. Mesh files live under `modelnet40_auto_aligned/` and are referenced by the `object_path` column (without that prefix).65 66## Load metadata with 🤗 Datasets67 68```python69from datasets import load_dataset70 71ds = load_dataset("naderalfares/ModelNet40_Auto_aligned")72 73print(ds)74# DatasetDict({75#     train: Dataset({ features: ['object_id', 'class', 'split', 'object_path', '__index_level_0__'], num_rows: 9843 })76#     test:  Dataset({ features: ['object_id', 'class', 'split', 'object_path', '__index_level_0__'], num_rows: 2468 })77# })78 79row = ds["train"][0]80print(row)81# {'object_id': 'airplane_0001', 'class': 'airplane', 'split': 'train',82#  'object_path': 'airplane/train/airplane_0001.off', '__index_level_0__': 100}83```84 85## Download a mesh file86 87```python88from huggingface_hub import hf_hub_download89 90repo_id = "naderalfares/ModelNet40_Auto_aligned"91row = ds["train"][0]92 93mesh_path = hf_hub_download(94    repo_id=repo_id,95    repo_type="dataset",96    filename=f"modelnet40_auto_aligned/{row['object_path']}",97)98print(mesh_path)  # local path to airplane_0001.off99```100 101## Citation102 103If you use this dataset, please cite the original ModelNet paper and the auto-alignment work:104 105```bibtex106@inproceedings{wu20153d,107  title={3D ShapeNets: A Deep Representation for Volumetric Shapes},108  author={Wu, Zhirong and Song, Shuran and Khademi, Adarsh and Zhao, Tian and others},109  booktitle={CVPR},110  year={2015}111}112```113 114## License115 116MIT117