Research on Painting Multi-Style Transfer Based on Perceptual Loss

Liucan Zhou

2024

Abstract

The art of various paintings has always been a reflection of people’s constant pursuit of aesthetics. Style transfer technology integrates various painting styles with image content, bringing this pursuit into people’s daily lives. Moreover, the development of deep learning technology and CNN application, style transfer technology has made significant breakthroughs. This paper mainly is about the method of style transfer based on CNN technology. On this topic, the Visual Geometry Group-19 (VGG-19) model is mainly used. By inputting preprocessed images into the VGG-19 model, it aids in isolating the style and content characteristics of images, which in turn facilitates the optimization of the perceptual loss function, and then uses the extracted perceptual loss function to implement a feedforward network for image transformation. This method can achieve satisfactory results and effectively completes the deep fusion of images with the desired style. This not only enhances the practicality of style transfer technology but also provides a new way for the pursuit of aesthetics.

Paper Citation

in Harvard Style

Zhou L. (2024). Research on Painting Multi-Style Transfer Based on Perceptual Loss. In Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML; ISBN 978-989-758-754-2, SciTePress, pages 401-405. DOI: 10.5220/0013524800004619

in Bibtex Style

@conference{daml24,
author={Liucan Zhou},
title={Research on Painting Multi-Style Transfer Based on Perceptual Loss},
booktitle={Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML},
year={2024},
pages={401-405},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013524800004619},
isbn={978-989-758-754-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Analysis and Machine Learning - Volume 1: DAML
TI - Research on Painting Multi-Style Transfer Based on Perceptual Loss
SN - 978-989-758-754-2
AU - Zhou L.
PY - 2024
SP - 401
EP - 405
DO - 10.5220/0013524800004619
PB - SciTePress