Academic Bibliography

 Search 200 years of publications by Ghent University researchers.
Advanced search
2 files | 6.54 MB
Add to list
  1. Search results
  2. CenDerNet : center and curvature repr...

CenDerNet : center and curvature representations for render-and-compare 6D pose estimation

Peter De Roovere (UGent) , Rembert Daems (UGent) , Jonathan Croenen, Taoufik Bourgana, Joris de Hoog and Francis wyffels (UGent)
(2023) Computer Vision : ECCV 2022 Workshops : proceedings, part VIII. In Lecture notes in computer science 13808. p.97-111
Author
Organization
Project
Abstract
We introduce CenDerNet, a framework for 6D pose estimation from multi-view images based on center and curvature representations. Finding precise poses for reflective, textureless objects is a key challenge for industrial robotics. Our approach consists of three stages: First, a fully convolutional neural network predicts center and curvature heatmaps for each view; Second, center heatmaps are used to detect object instances and find their 3D centers; Third, 6D object poses are estimated using 3D centers and curvature heatmaps. By jointly optimizing poses across views using a render-and-compare approach, our method naturally handles occlusions and object symmetries. We show that CenDerNet outperforms previous methods on two industry-relevant datasets: DIMO and T-LESS .
Keywords
Object detection, 6D object pose estimation, Industrial robotics, Render-and-Compare, Curvature maps

Downloads

  • Download
    DS606 acc.pdf
    • full text (Author's original)
    • |
    • open access
    • |
    • PDF
    • |
    • 3.99 MB
  • Download
    (...).pdf
    • full text (Published version)
    • |
    • UGent only
    • |
    • PDF
    • |
    • 2.55 MB

Citation

Please use this url to cite or link to this publication:

MLA
De Roovere, Peter, et al. “CenDerNet : Center and Curvature Representations for Render-and-Compare 6D Pose Estimation.” Computer Vision : ECCV 2022 Workshops : Proceedings, Part VIII, vol. 13808, Springer, 2023, pp. 97–111, doi:10.1007/978-3-031-25085-9_6.
APA
De Roovere, P., Daems, R., Croenen, J., Bourgana, T., de Hoog, J., & wyffels, F. (2023). CenDerNet : center and curvature representations for render-and-compare 6D pose estimation. Computer Vision : ECCV 2022 Workshops : Proceedings, Part VIII, 13808, 97–111. https://doi.org/10.1007/978-3-031-25085-9_6
Chicago author-date
De Roovere, Peter, Rembert Daems, Jonathan Croenen, Taoufik Bourgana, Joris de Hoog, and Francis wyffels. 2023. “CenDerNet : Center and Curvature Representations for Render-and-Compare 6D Pose Estimation.” In Computer Vision : ECCV 2022 Workshops : Proceedings, Part VIII, 13808:97–111. Cham: Springer. https://doi.org/10.1007/978-3-031-25085-9_6.
Chicago author-date (all authors)
De Roovere, Peter, Rembert Daems, Jonathan Croenen, Taoufik Bourgana, Joris de Hoog, and Francis wyffels. 2023. “CenDerNet : Center and Curvature Representations for Render-and-Compare 6D Pose Estimation.” In Computer Vision : ECCV 2022 Workshops : Proceedings, Part VIII, 13808:97–111. Cham: Springer. doi:10.1007/978-3-031-25085-9_6.
Vancouver
1.
De Roovere P, Daems R, Croenen J, Bourgana T, de Hoog J, wyffels F. CenDerNet : center and curvature representations for render-and-compare 6D pose estimation. In: Computer Vision : ECCV 2022 Workshops : proceedings, part VIII. Cham: Springer; 2023. p. 97–111.
IEEE
[1]
P. De Roovere, R. Daems, J. Croenen, T. Bourgana, J. de Hoog, and F. wyffels, “CenDerNet : center and curvature representations for render-and-compare 6D pose estimation,” in Computer Vision : ECCV 2022 Workshops : proceedings, part VIII, Tel Aviv, Israel, 2023, vol. 13808, pp. 97–111.
@inproceedings{01GSSRRT111G3T3MQE77BSZ7MR,
  abstract     = {{We introduce CenDerNet, a framework for 6D pose estimation from multi-view images based on center and curvature representations. Finding precise poses for reflective, textureless objects is a key challenge for industrial robotics. Our approach consists of three stages: First, a fully convolutional neural network predicts center and curvature heatmaps for each view; Second, center heatmaps are used to detect object instances and find their 3D centers; Third, 6D object poses are estimated using 3D centers and curvature heatmaps. By jointly optimizing poses across views using a render-and-compare approach, our method naturally handles occlusions and object symmetries. We show that CenDerNet outperforms previous methods on two industry-relevant datasets: DIMO and T-LESS .}},
  author       = {{De Roovere, Peter and Daems, Rembert and Croenen, Jonathan and Bourgana, Taoufik and de Hoog, Joris and wyffels, Francis}},
  booktitle    = {{Computer Vision : ECCV 2022 Workshops : proceedings, part VIII}},
  isbn         = {{9783031250842}},
  issn         = {{0302-9743}},
  keywords     = {{Object detection,6D object pose estimation,Industrial robotics,Render-and-Compare,Curvature maps}},
  language     = {{eng}},
  location     = {{Tel Aviv, Israel}},
  pages        = {{97--111}},
  publisher    = {{Springer}},
  title        = {{CenDerNet : center and curvature representations for render-and-compare 6D pose estimation}},
  url          = {{http://doi.org/10.1007/978-3-031-25085-9_6}},
  volume       = {{13808}},
  year         = {{2023}},
}
Altmetric
View in Altmetric