Shape Transformation with CycleGAN Using an Automobile as an Example

Akira Nakajima, Hiroyuki Kobayashi

2023

Abstract

AI technology has developed remarkably in recent years, and AI-based image generation tools have spread rapidly. CycleGAN is one of the image generation AIs and specializes in image style transformation, and has the problem of being able to change colors and patterns but not shapes. The reason may be that the model considers the background as a part of the conversion target, which can be solved by removing the background. In this study, the number of backgrounds is limited to a certain number, and CycleGAN is used for shape transformation.The evaluation is done by comparing the result of this experiment with the image transformation when the input is an image with the background removed.Comparison of the proposed and conventional methods showed comparable results.

Paper Citation

in Harvard Style

Nakajima A. and Kobayashi H. (2023). Shape Transformation with CycleGAN Using an Automobile as an Example. In Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO; ISBN 978-989-758-670-5, SciTePress, pages 736-739. DOI: 10.5220/0012233100003543

in Bibtex Style

@conference{icinco23,
author={Akira Nakajima and Hiroyuki Kobayashi},
title={Shape Transformation with CycleGAN Using an Automobile as an Example},
booktitle={Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO},
year={2023},
pages={736-739},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012233100003543},
isbn={978-989-758-670-5},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 20th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO
TI - Shape Transformation with CycleGAN Using an Automobile as an Example
SN - 978-989-758-670-5
AU - Nakajima A.
AU - Kobayashi H.
PY - 2023
SP - 736
EP - 739
DO - 10.5220/0012233100003543
PB - SciTePress