Kaputt: A Large-Scale Dataset for Visual Defect Detection
Paper • 2510.05903 • Published
Datasets:
Access the Kaputt Defect Dataset (KDD)
Access to the Kaputt Defect Dataset is granted after you accept its Terms of Use. By requesting access you agree to share your Hugging Face username and email address with the dataset authors.
The Kaputt Defect Dataset (KDD) is subject to the Creative Commons CC BY-NC-ND 4.0 license, with additional Superseding Terms that take precedence where they conflict with the Creative Commons license. The full, binding text is in the LICENSE file in this repository. In summary: you agree to use the KDD solely for education and scientific research in the field of computer vision; it is provided "as is" with no warranties; you are solely responsible for lawful use of any third-party intellectual property it may contain; disputes are adjudicated in King County, Washington. You must read and accept the full Terms of Use (LICENSE) before access is granted.
Log in or Sign Up to review the conditions and access this dataset content.
Kaputt: A Large-Scale Dataset for Visual Defect Detection
Abstract
We present a novel large-scale dataset for defect detection in a logistics setting. Recent work on industrial anomaly detection has primarily focused on manufacturing scenarios with highly controlled poses and a limited number of object categories. Existing benchmarks like MVTec-AD (Bergmann et al., 2021) and VisA (Zou et al., 2022) have reached saturation, with state-of-the-art methods achieving up to 99.9% AUROC scores. In contrast to manufacturing, anomaly detection in retail logistics faces new challenges, particularly in the diversity and variability of object pose and appearance. Leading anomaly detection methods fall short when applied to this new setting. To bridge this gap, we introduce a new benchmark that overcomes the current limitations of existing datasets. With over 230,000 images (and more than 29,000 defective instances), it is 40 times larger than MVTec and contains more than 48,000 distinct objects. To validate the difficulty of the problem, we conduct an extensive evaluation of multiple state-of-the-art anomaly detection methods, demonstrating that they do not surpass 56.96% AUROC on our dataset. Further qualitative analysis confirms that existing methods struggle to leverage normal samples under heavy pose and appearance variation. With our large-scale dataset, we set a new benchmark and encourage future research towards solving this challenging problem in retail logistics anomaly detection. The dataset is available for download under https://www.kaputt-dataset.com.
- Venue: IEEE/CVF International Conference on Computer Vision (ICCV), October 2025, Honolulu, Hawaii, USA
- Paper: https://arxiv.org/abs/2510.05903
- Project page: https://www.kaputt-dataset.com
Authors
Sebastian Höfer¹, Dorian Henning¹, Artemij Amiranashvili¹, Douglas Morrison¹, Mariliza Tzes¹, Ingmar Posner¹˒², Marc Matvienko¹, Alessandro Rennola¹, Anton Milan¹.
¹ Amazon ² University of Oxford
 |
 |
 |
Citation
@inproceedings{kaputt2025,
title = {Kaputt: A Large-Scale Dataset for Visual Defect Detection},
author = {H{\"o}fer, Sebastian and Henning, Dorian and Amiranashvili, Artemij and Morrison, Douglas and Tzes, Mariliza and Posner, Ingmar and Matvienko, Marc and Rennola, Alessandro and Milan, Anton},
booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
month = {October},
year = {2025},
address = {Honolulu, Hawaii, USA},
publisher = {IEEE}
}
License
This dataset is released under CC BY-NC-ND 4.0 with additional Superseding Terms that take precedence where they conflict with the Creative Commons license. See the LICENSE file for the full, binding text.
Use is permitted solely for education and scientific research in the field of computer vision. Per Clause 7 of the Terms, you may not remove the terms from the dataset.
Acknowledgements
We thank our collaborators in Amazon's operations, hardware and software engineering, as well as our annotation teams. Their invaluable contributions to hardware development, software implementation, data collection, and labeling efforts were essential to the success of this work.
- Downloads last month
- 14