AesFA: An Aesthetic Feature-Aware Arbitrary Neural Style Transfer

arXiv:2312.05928v3 Announce Type: replace
Abstract: Neural style transfer (NST) has evolved significantly in recent years. Yet, despite its rapid progress and advancement, existing NST methods either struggle to transfer aesthetic information from a style effectively or suffer from high computational costs and inefficiencies in feature disentanglement due to using pre-trained models. This work proposes a lightweight but effective model, AesFA -- Aesthetic Feature-Aware NST. The primary idea is to decompose the image via its frequencies to better disentangle aesthetic styles …

arxiv cs.ai cs.cv feature style style transfer transfer type