---
license: mit
tags:
- industrial
- anomaly-detection
- multi-view
- object-centric
- parts
---

# Dataset Card for Dataset Name

<!-- Provide a quick summary of the dataset. -->
This dataset contains 7 industrial parts taken from a 360-degree all-around view with one axis of rotation.
It contains a train set of all-around images and test sets with ground truth anomaly masks.

**Paper:** [[3D Gaussian Reference Parts for Robust Free-Viewpoint Visual Inspection](https://ieeexplore.ieee.org/document/11424398)](https://ieeexplore.ieee.org/document/11424398)

**BibTeX entry and citation info**
```bibtex
@ARTICLE{11424398,
  author={Ito, Kenta and Ueda, Shiori and Mori, Shohei and Sugano, Junichi and Adachi, Hideyuki and Saito, Hideo},
  journal={IEEE Access}, 
  title={3D Gaussian Reference Parts for Robust Free-Viewpoint Visual Inspection}, 
  year={2026},
  volume={14},
  number={},
  pages={37885-37896},
  keywords={Three-dimensional displays;Inspection;Pipelines;Feature extraction;Cameras;Pose estimation;Visualization;Defect detection;Point cloud compression;Image reconstruction;Digital twin;3D reconstruction;pose estimation;template matching;visual inspection},
  doi={10.1109/ACCESS.2026.3671877}
  }
```