Synthetic Ground Truth for Presegmentation of Known Objects for Effortless Pose Estimation

Frederik Haarslev, William Kristian Juel, Norbert Krüger, Leon Bodenhagen

2020

Abstract

We present a method for generating synthetic ground truth for training segmentation networks for presegmenting point clouds in pose estimation problems. Our method replaces global pose estimation algorithms such as RANSAC which requires manual fine-tuning with a robust CNN, without having to hand-label segmentation masks for the given object. The data is generated by blending cropped images of the objects with arbitrary backgrounds. We test the method in two scenarios, and show that networks trained on the generated data segments the objects with high accuracy, allowing them to be used in a pose estimation pipeline.

Download


Paper Citation


in Harvard Style

Haarslev F., Juel W., Krüger N. and Bodenhagen L. (2020). Synthetic Ground Truth for Presegmentation of Known Objects for Effortless Pose Estimation. In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2020) - Volume 4: VISAPP; ISBN 978-989-758-402-2, SciTePress, pages 482-489. DOI: 10.5220/0009163904820489


in Bibtex Style

@conference{visapp20,
author={Frederik Haarslev and William Kristian Juel and Norbert Krüger and Leon Bodenhagen},
title={Synthetic Ground Truth for Presegmentation of Known Objects for Effortless Pose Estimation},
booktitle={Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2020) - Volume 4: VISAPP},
year={2020},
pages={482-489},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0009163904820489},
isbn={978-989-758-402-2},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2020) - Volume 4: VISAPP
TI - Synthetic Ground Truth for Presegmentation of Known Objects for Effortless Pose Estimation
SN - 978-989-758-402-2
AU - Haarslev F.
AU - Juel W.
AU - Krüger N.
AU - Bodenhagen L.
PY - 2020
SP - 482
EP - 489
DO - 10.5220/0009163904820489
PB - SciTePress