Dataset Credits
Some images, point clouds, meshes and videos shown in this documentation, and some example inputs used by the Telekinesis examples, are derived from the public research datasets below. We thank their authors.
The data is redistributed under each dataset's licence, with the changes noted. These licences apply to the data only, not to the Telekinesis software or documentation text. A file-by-file list of example inputs is in THIRD_PARTY_DATA.md in the examples repository.
Changes made by Telekinesis (all datasets): images were cropped and/or resized, and skill results were rendered on top of them. Further changes are listed per dataset.
XYZ Industrial Bin-Picking Dataset (XYZ-IBD)
- Authors: Junwen Huang, Jiaqi Hu, Peter KT Yu, Slobodan Ilic, Martin Sundermeyer, Benjamin Busam (TUM, MCML, Google, XYZ Robotics)
- Citation: J. Huang et al., "XYZ-IBD: Benchmarking Robust 6D Object Pose Estimation under Real-World Industrial Complexity", arXiv:2506.00599, 2025.
- Source: xyz-ibd.github.io
- Licence: CC BY-NC-SA 4.0. Used by Telekinesis with the written permission of the authors.
- Further changes: depth maps back-projected to point clouds, ground-truth masks merged or cropped, 16-bit depth holes inpainted; two images contain simulated noise.
ARMBench
- Authors: Amazon Robotics (C. Mitash, F. Wang, S. Lu, V. Terhuja, T. Garaas, F. Polido, M. Nambi)
- Citation: C. Mitash et al., "ARMBench: An Object-centric Benchmark Dataset for Robotic Manipulation", ICRA 2023, arXiv:2303.16382.
- Source: armbench.com
- Licence: CC BY 4.0
Visual Anomaly (VisA)
- Authors: Amazon (Y. Zou, J. Jeong, L. Pemula, D. Zhang, O. Dabeer)
- Citation: Y. Zou et al., "SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation", ECCV 2022, arXiv:2207.14315.
- Source: registry.opendata.aws/visa
- Licence: CC BY 4.0
LOCO: Logistics Objects in Context
- Authors: C. Mayershofer, D.-M. Holm, B. Molter, J. Fottner (Technical University of Munich, fml)
- Citation: C. Mayershofer et al., "LOCO: Logistics Objects in Context", ICMLA 2020.
- Source: github.com/tum-fml/loco
- Licence: CC0 1.0 (public domain; credited as a courtesy)
- Further changes: one unblurred face was pixelated by Telekinesis.
FabriD: Fabric Defect Detection Dataset
- Authors: Sabbir Ahmed, Ismam Ahmed, Nishat Mahzabin, Tahmina Islam (University of Information Technology and Sciences, Dhaka)
- Citation: S. Ahmed et al., "FabriD: Fabric Defect Detection Dataset", Mendeley Data, V3, doi:10.17632/vncdwps24r.3.
- Licence: CC BY 4.0
MATWI: A Multimodal Automatic Tool Wear Inspection dataset
- Authors: Lars De Pauw, Tom Jacobs, Toon Goedemé (KU Leuven)
- Citation: L. De Pauw, T. Jacobs, T. Goedemé, "Replication Data for: MATWI", KU Leuven RDR, 2025, doi:10.48804/GK6LHH.
- Licence: CC BY-SA 4.0. The modified MATWI images shown here are themselves licensed under CC BY-SA 4.0.
T-LESS
- Authors: T. Hodaň, P. Haluza, Š. Obdržálek, J. Matas, M. Lourakis, X. Zabulis (Czech Technical University in Prague)
- Citation: T. Hodaň et al., "T-LESS: An RGB-D Dataset for 6D Pose Estimation of Texture-less Objects", WACV 2017.
- Source: cmp.felk.cvut.cz/t-less
- Licence: CC BY 4.0
- Further changes: depth back-projected to point clouds; CAD meshes sliced and converted.
Contact
If you are a dataset author and would like a credit corrected, please contact suman.pal@telekinesis.ai.