IntroStyle: Training-Free Introspective Style Attribution using Diffusion Features

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Kumar, Anand, Mu, Jiteng, Vasconcelos, Nuno
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918114906603520
author Kumar, Anand
Mu, Jiteng
Vasconcelos, Nuno
author_facet Kumar, Anand
Mu, Jiteng
Vasconcelos, Nuno
contents Text-to-image (T2I) models have recently gained widespread adoption. This has spurred concerns about safeguarding intellectual property rights and an increasing demand for mechanisms that prevent the generation of specific artistic styles. Existing methods for style extraction typically necessitate the collection of custom datasets and the training of specialized models. This, however, is resource-intensive, time-consuming, and often impractical for real-time applications. We present a novel, training-free framework to solve the style attribution problem, using the features produced by a diffusion model alone, without any external modules or retraining. This is denoted as Introspective Style attribution (IntroStyle) and is shown to have superior performance to state-of-the-art models for style attribution. We also introduce a synthetic dataset of Artistic Style Split (ArtSplit) to isolate artistic style and evaluate fine-grained style attribution performance. Our experimental results on WikiArt and DomainNet datasets show that \ours is robust to the dynamic nature of artistic styles, outperforming existing methods by a wide margin.
format Preprint
id arxiv_https___arxiv_org_abs_2412_14432
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle IntroStyle: Training-Free Introspective Style Attribution using Diffusion Features
Kumar, Anand
Mu, Jiteng
Vasconcelos, Nuno
Computer Vision and Pattern Recognition
Image and Video Processing
Text-to-image (T2I) models have recently gained widespread adoption. This has spurred concerns about safeguarding intellectual property rights and an increasing demand for mechanisms that prevent the generation of specific artistic styles. Existing methods for style extraction typically necessitate the collection of custom datasets and the training of specialized models. This, however, is resource-intensive, time-consuming, and often impractical for real-time applications. We present a novel, training-free framework to solve the style attribution problem, using the features produced by a diffusion model alone, without any external modules or retraining. This is denoted as Introspective Style attribution (IntroStyle) and is shown to have superior performance to state-of-the-art models for style attribution. We also introduce a synthetic dataset of Artistic Style Split (ArtSplit) to isolate artistic style and evaluate fine-grained style attribution performance. Our experimental results on WikiArt and DomainNet datasets show that \ours is robust to the dynamic nature of artistic styles, outperforming existing methods by a wide margin.
title IntroStyle: Training-Free Introspective Style Attribution using Diffusion Features
topic Computer Vision and Pattern Recognition
Image and Video Processing
url https://arxiv.org/abs/2412.14432