IKEA 3D Assembly Dataset
The IKEA 3D Assembly Dataset is provided by Inter IKEA Systems B.V. to encourage research related to assembly instructions. It contains 3D models of five IKEA products that have been prepared so that each part a user interacts with during assembly is easily manipulable in 3D.
Dataset description and usage context
IKEA 3D Assembly Dataset Overview
Dataset Description
The IKEA 3D Assembly Dataset, provided by Inter IKEA Systems B.V., aims to promote research on assembly instructions. The dataset includes 3D files for five IKEA products; in the final assembled state, all parts that users interact with during assembly are readily operable in 3D.
Key Features
- Contains 3D models of five IKEA products, with all user‑interacted parts included.
- Basic color/material assignment for product components.
- Corresponding assembly instruction documents for the current version.
Products in the Dataset
| Image | Product Description |
|---|---|
![]() | LACK 55x55 cm, white – side table |
![]() | EKET 35×25×35 cm, white – small cabinet |
![]() | EKET 70×35×35 cm, white – cabinet with two drawers |
![]() | BEKVÄM black – step stool |
![]() | DALFRED black – bar stool |
Dataset Formats
| File Type | Description | Unit |
|---|---|---|
| .glb/.glTF | Final assembled state with materials | meters |
| .obj | Final assembled state with basic colors only | millimeters |
| Assembly instructions | – |
Versioning
Version numbers follow the major.minor scheme. The minor version indicates non‑breaking changes (e.g., format extensions). The major version denotes breaking changes such as adding or removing models or altering material definitions.
License
The dataset is released under CC BY‑NC‑SA 4.0, for non‑commercial use only, to foster research on furniture assembly.
Citation
When using this dataset, please cite:
@misc{su2021ikea,
title={IKEA Object State Dataset: A 6DoF object pose estimation dataset and benchmark for multi‑state assembly objects},
author={Yongzhi Su and Mingxin Liu and Jason Rambach and Antonia Pehrson and Anton Berg and Didier Stricker},
year={2021},
eprint={2111.08614},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
Pair the dataset with AI analysis and content workflows.
Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.




