thingi10k 1.1.3

Thingi10k dataset.

Thingi10K Dataset

Thingi10K is a large scale 3D dataset created to study the variety, complexity and quality of real-world 3D printing models. We analyze every mesh of all things featured on Thingiverse.com between Sept. 16, 2009 and Nov. 15, 2015. On this site, we hope to share our findings with you.

In a nutshell, Thingi10K contains...

• 10,000 models
• 4,892 tags
• 2,011 things
• 1,083 designers
• 72 categories
• 10 open source licenses
• 7+ years span
• 99.6% .stl files
• 50% non-solid
• 45% with self-intersections
• 31% with coplanar self-intersections
• 26% with multiple components
• 22% non-manifold
• 16% with degenerate faces
• 14% non-PWN
• 11% topologically open
• 10% non-oriented

Thingi10K is created by Qingnan Zhou and Alec Jacobson.

Raw dataset

You can download the raw dataset at Hugging Face.

One can also obtain the dataset via the thingi10k Python package. It contains both geometric and contextual data extracted from the raw dataset, and provides a convenient API to access and filter the dataset.

Installation

pip install thingi10k

Usage

import thingi10k

thingi10k.init() # Initial download of the dataset

# Iterate over the entire dataset
for entry in thingi10k.dataset():
    file_id = entry['file_id']

    # Check contextual data
    author = entry['author']
    license = entry['license']

    # Load actual geometry
    vertices, facets = thingi10k.load_file(entry['file_path'])

Variants

Thingi10K provides two variants of the dataset: npz and raw.

• npz variant contains the geometry (vertex and facet arrays) in NumPy arrays. It is faster to download and no mesh parsing is necessary.
• raw variant contains the raw mesh files (STL, OBJ, etc.) in their original format. It is slower to download and requires parsing to extract geometric data.

By default, thingi10k.init() will download the npz variant. To download the raw variant:

thingi10k.init(variant='raw')

Caching the dataset

By default, thingi10k.init() will cache the dataset in a local directory. Any subsequent calls to thingi10k.init() will use the cached dataset and incur no additional download cost. The cache directory can be explicitly specified by user:

thingi10k.init(cache_dir="path/to/.thingi10k")

To force a re-download of the dataset:

thingi10k.init(force_redownload=True)

Filtering the dataset

Thingi10K supports advanced filtering of the dataset based on geometric and contextual properties. See help(thingi10k.dataset) for a full list of supported filters. Here are some examples:

# Filter by closed models
closed_models = thingi10k.dataset(closed=True)

# Filter by models without self-intersections
no_self_intersections = thingi10k.dataset(self_intersecting=False)

# Filter by the number of vertices (<100)
small_models = thingi10k.dataset(num_vertices=(None, 100))

# Models with a single component and is manifold
single_manifold_models = thingi10k.dataset(num_components=1, manifold=True)

# Models licensed under the Creative Commons license
cc_models = thingi10k.dataset(license='creative commons')

License

The source code for organizing and filtering the Thingi10K dataset is licensed under the Apache License, Version 2.0. Each "thing" in the dataset is licensed under different licenses. Please refer to the license field associated with each entry in the dataset.

Errata

The following models are known to be "corrupt." However, we decide to still include them in our dataset in order to faithfully reflect mesh qualities on Thingiverse.

• Model 49911 is truncated (ASCII STL).
• Model 74463 is empty.
• Model 286163 is empty.
• Model 81313 contains NURBS curves and surfaces instead of polygonal faces, which may not be supported by many OBJ parsers.
• Model 77942 is corrupt (binary STL).

Acknowledgements

This project is funded in part by NSF grants CMMI-11-29917, IIS-14-09286, and IIS-17257.

We thank Marcel Campen, Chelsea Tymms, and Julian Panetta for early feedback and proofreading. We also thank Neil Dickson for pointing out corrupt models, and Nick Sharp for pointing out bugs in download script. Lastly, we thank Silvia Sellán and Yun-Chun Chen for discussion and suggestion on hosting the dataset.

Cite us

@article{Thingi10K,
  title={Thingi10K: A Dataset of 10,000 3D-Printing Models},
  author={Zhou, Qingnan and Jacobson, Alec},
  journal={arXiv preprint arXiv:1605.04797},
  year={2016}
}