Acknowledge license to accept the repository
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
By using data from this repo, users agree to the corresponding licensing of the original data.
To request access to ArtiAtlas, you will need to provide your full name (please provide both your first and last name), the name of your advisor or the principal investigator (PI) of your lab (in the PI/Advisor field), and the school or company that you are affiliated with (the Affiliation field). After requesting access to this repo, you will be considered for access approval.
Log in or Sign Up to review the conditions and access this dataset content.
ArtiAtlas: A Large-Scale Dataset of Annotated Articulated 3D Objects
Articulated 3D objects across 173 categories, annotated with rigid functional links, a rooted kinematic tree over those links, and per-joint parameters — joint type, axis, pivot and motion range. Every object exports to a URDF.
From ArtiAtlas: A Holistic Ecosystem for Diverse 3D Articulation Modeling (SIGGRAPH Asia 2026). Code, benchmark and model: Seed3D/ArtiAtlas.
This is a first release. The data is still being verified and improved, and further assets are being prepared — including objects whose upstream licensing means we will publish source links and annotations rather than geometry. We will add them as they are ready, and the layout below may still change between drops. If an object looks wrong, or an annotation disagrees with its geometry, please open a discussion on this repository — that feedback is what makes the next drop better.
Data Structure
data/
{oid}/ # 5-digit zero-padded object identifier
mobility.urdf # one link per rigid part, one joint per articulation.
# Inertial properties are placeholders.
relation.json # kinematic tree over the links
motion.json # per-joint motion type and range
category.json # taxonomy labels
meta.json # upstream dataset the object came from, and its id there
textured_objs/ # part_NNN.obj + part_NNN.mtl
images/ # texture maps, referenced as ../images/<file> from the
# .mtl files; empty for an untextured object
manifest.json
Download
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="Moon-bow/ArtiAtlas",
repo_type="dataset",
local_dir="artiatlas",
)
The repository holds the tree above as one tar.gz, with manifest.json beside it indexing
every object. Verify and unpack it:
cd artiatlas
sha256sum -c artiatlas-research-sources.tar.gz.sha256
tar xzf artiatlas-research-sources.tar.gz # -> data/
Related
- Artisan — the model trained on this dataset
Removal Requests
If there is any license issue, please contact us and we will remove the data.
Citation
@inproceedings{artiatlas2026,
title = {ArtiAtlas: A Holistic Ecosystem for Diverse 3D Articulation Modeling},
author = {Lu, Junzhe and Lin, Jing and Cao, Ziang and Wang, Haoqian and Liu, Ziwei and Zhang, Jianfeng},
booktitle = {SIGGRAPH Asia 2026 Conference Papers},
year = {2026}
}
- Downloads last month
- 14