SteelDS: A High-Resolution Video Dataset of E40 Steel Scrap for Object Detection and Instance Segmentation

The SteelDS dataset is a high-resolution video and image repository optimized for instance segmentation, object detection, object tracking, and material classification tasks within industrial metal recycling automation. The dataset specifically targets shredded E40-grade steel and copper scrap particles under simulated industrial conveyor belt conditions.

Please cite the following paper if you use this dataset in your research:

Melanie Neubauer, Christian Rauch, Gerald Koinig, Alexia Tischberger-Aldrian, Roland Pomberger, Elmar Rueckert. "SteelDS: A High-Resolution Video Dataset of E40 Steel Scrap for Object Detection and Instance Segmentation", arXiv:2605.26682 [cs.RO], 2026.

BibTeX:
@misc{neubauer2026steelds,
  author        = {Neubauer, Melanie and Rauch, Christian and Koinig, Gerald and Tischberger-Aldrian, Alexia and Pomberger, Roland and Rueckert, Elmar},
  title         = {{SteelDS}: A High-Resolution Video Dataset of {E40} Steel Scrap for Object Detection and Instance Segmentation},
  year          = {2026},
  eprint        = {2605.26682},
  archivePrefix = {arXiv},
  primaryClass  = {cs.RO},
  doi           = {10.48550/arXiv.2605.26682},
  url           = {https://arxiv.org/abs/2605.26682}
}