README.md

---
task_categories:
- other
tags:
- 3d
- articulated-objects
- mesh
pretty_name: Articulate-100
size_categories:
- 10K<n<100K
---

# Articulate-100

Released data for

**ArtMesh: Part-Aware Articulated Mesh Fields with Motion-Consistent Dynamics**

[Project Website](https://artmesh2026.github.io/artmesh-site/) | [arXiv Paper](https://arxiv.org/abs/2605.16582)

## Overview

Articulate-100 is a collection of articulated object instances spanning common
household and workspace categories. Each archive contains the assets for one
object instance.

Categories represented include boxes, eyeglasses, faucets, folding chairs,
knives, laptops, ovens, pens, pliers, refrigerators, storage furniture,
suitcases, tables, toilets, trash cans, and windows.

## Structure

Each object is distributed as a single `.tar` archive named
`<Category>_<InstanceID>.tar`. Extracting an archive yields a directory
`<Category>_<InstanceID>/` containing the assets for that instance.

```
Articulate-100/
├── Box_100129.tar
├── Box_102379.tar
├── Eyeglasses_101293.tar
├── ...
```

## Usage

Download the full dataset:

```python
from huggingface_hub import snapshot_download

local_dir = snapshot_download(
    repo_id="ArtMesh/Articulate-100",
    repo_type="dataset",
)
```

Download a single object:

```python
from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="ArtMesh/Articulate-100",
    repo_type="dataset",
    filename="Box_100129.tar",
)
```

Extract:

```bash
tar -xf Box_100129.tar
```

## Source

The object assets are derived from
[PartNet-Mobility](https://sapien.ucsd.edu/browse), part of the SAPIEN project.
Please consult the original dataset for terms governing use of the underlying
assets.