AbSynth: Using Abstract Image Synthesis for Synthetic Training

Dominik Penk, Dominik Penk, Maik Horn, Christoph Strohmeyer, Bernhard Egger, Marc Stamminger, Frank Bauer

2024

Abstract

We present a novel pipeline for training neural networks to tackle geometry-induced vision tasks, relying solely on synthetic training images generated from (geometric) CAD models of the objects under consideration. Instead of aiming for photorealistic renderings, our approach maps both synthetic and real-world data onto a common abstract image space reducing the domain gap. We demonstrate that this projection can be decoupled from the downstream task, making our method an easy drop-in solution for a variety of applications. In this paper, we use line images as our chosen abstract image representation due to their ability to capture geometric properties effectively. We introduce an efficient training data synthesis method, that generates images tailored for transformation into a line representation. Additionally, we explore how the use of sparse line images opens up new possibilities for augmenting the dataset, enhancing the overall robustness of the downstream models. Finally, we provide an evaluation of our pipeline and augmentation techniques across a range of vision tasks and state-of-the-art models, showcasing their effectiveness and potential for practical applications.

Paper Citation

in Harvard Style

Penk D., Horn M., Strohmeyer C., Egger B., Stamminger M. and Bauer F. (2024). AbSynth: Using Abstract Image Synthesis for Synthetic Training. In Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP; ISBN 978-989-758-679-8, SciTePress, pages 718-729. DOI: 10.5220/0012431400003660

in Bibtex Style

@conference{visapp24,
author={Dominik Penk and Maik Horn and Christoph Strohmeyer and Bernhard Egger and Marc Stamminger and Frank Bauer},
title={AbSynth: Using Abstract Image Synthesis for Synthetic Training},
booktitle={Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP},
year={2024},
pages={718-729},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012431400003660},
isbn={978-989-758-679-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 2: VISAPP
TI - AbSynth: Using Abstract Image Synthesis for Synthetic Training
SN - 978-989-758-679-8
AU - Penk D.
AU - Horn M.
AU - Strohmeyer C.
AU - Egger B.
AU - Stamminger M.
AU - Bauer F.
PY - 2024
SP - 718
EP - 729
DO - 10.5220/0012431400003660
PB - SciTePress