Keep It Simple: Local Search-based Latent Space Editing

Andreas Meißner, Andreas Meißner, Andreas Fröhlich, Michaela Geierhos

2022

Abstract

Semantic image editing allows users to selectively change entire image attributes in a controlled manner with just a few clicks. Most approaches use a generative adversarial network (GAN) for this task to learn an appropriate latent space representation and attribute-specific transformations. While earlier approaches often suffer from entangled attribute manipulations, newer ones improve on this aspect by using separate specialized networks for attribute extraction. Iterative optimization algorithms based on backpropagation constitute a possible approach to find attribute vectors with little entanglement. However, this requires a large amount of GPU memory, training instabilities can occur, and the used models have to be differentiable. To address these issues, we propose a local search-based approach for latent space editing. We show that it performs at the same level as previous algorithms and avoids these drawbacks.

Paper Citation

in Harvard Style

Meißner A., Fröhlich A. and Geierhos M. (2022). Keep It Simple: Local Search-based Latent Space Editing. In Proceedings of the 14th International Joint Conference on Computational Intelligence (IJCCI 2022) - Volume 1: NCTA; ISBN 978-989-758-611-8, SciTePress, pages 273-283. DOI: 10.5220/0011524700003332

in Bibtex Style

@conference{ncta22,
author={Andreas Meißner and Andreas Fröhlich and Michaela Geierhos},
title={Keep It Simple: Local Search-based Latent Space Editing},
booktitle={Proceedings of the 14th International Joint Conference on Computational Intelligence (IJCCI 2022) - Volume 1: NCTA},
year={2022},
pages={273-283},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011524700003332},
isbn={978-989-758-611-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 14th International Joint Conference on Computational Intelligence (IJCCI 2022) - Volume 1: NCTA
TI - Keep It Simple: Local Search-based Latent Space Editing
SN - 978-989-758-611-8
AU - Meißner A.
AU - Fröhlich A.
AU - Geierhos M.
PY - 2022
SP - 273
EP - 283
DO - 10.5220/0011524700003332
PB - SciTePress