IntereStyle: Encoding an Interest Region for Robust StyleGAN Inversion

Moon, Seungjun; Park, Gyeong-Moon

doi:10.1007/978-3-031-19784-0_27

Cited by 6 publications

(2 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To apply this extraordinary latent space to real images, several works [1,2,35,40,36] have bridged the relationship between real and fake latent spaces through a process known as GAN inversion. GAN inversion methods can typically be divided into optimization-based [1,2,10,34] and encoder-based [35,40,3,47,4,49,33]. While optimization-based models exhibit high inversion quality, they require numerous optimization steps for each input image [23], resulting in significant time consumption.…”

Section: Related Workmentioning

confidence: 99%

GaitEditer: Attribute Editing for Gait Representation Learning

Ye¹,

Ma²,

Chen³

et al. 2023

Preprint

View full text Add to dashboard Cite

Section: Related Workmentioning

confidence: 99%

GaitEditer: Attribute Editing for Gait Representation Learning

Ye¹,

Ma²,

Chen³

et al. 2023

Preprint

View full text Add to dashboard Cite

“…Encoder-based approaches (Richardson et al 2021;Tov et al 2021;Hu et al 2022;Mao et al 2022) map latent codes more quickly through a learnable encoder, with better editability but worse fidelity. To keep more details, (Wei et al 2022;Moon and Park 2022) complement latent codes with the differences between reconstructed and source images. Other works (Roich et al 2022;Dinh et al 2022) have attempted to fine-tune the generator.…”

Section: Introductionmentioning

confidence: 99%

Gradual Residuals Alignment: A Dual-Stream Framework for GAN Inversion and Image Attribute Editing

Li,

Huang,

Zhang

et al. 2024

AAAI

View full text Add to dashboard Cite

GAN-based image attribute editing firstly leverages GAN Inversion to project real images into the latent space of GAN and then manipulates corresponding latent codes. Recent inversion methods mainly utilize additional high-bit features to improve image details preservation, as low-bit codes cannot faithfully reconstruct source images, leading to the loss of details. However, during editing, existing works fail to accurately complement the lost details and suffer from poor editability. The main reason is they inject all the lost details indiscriminately at one time, which inherently induces the position and quantity of details to overfit source images, resulting in inconsistent content and artifacts in edited images. This work argues that details should be gradually injected into both the reconstruction and editing process in a multi-stage coarse-to-fine manner for better detail preservation and high editability. Therefore, a novel dual-stream framework is proposed to accurately complement details at each stage. The Reconstruction Stream is employed to embed coarse-to-fine lost details into residual features and then adaptively add them to the GAN generator. In the Editing Stream, residual features are accurately aligned by our Selective Attention mechanism and then injected into the editing process in a multi-stage manner. Extensive experiments have shown the superiority of our framework in both reconstruction accuracy and editing quality compared with existing methods.

show abstract

Multi-domain Information Fusion for Key-Points Guided GAN Inversion

Xu,

Qiu,

et al. 2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

IntereStyle: Encoding an Interest Region for Robust StyleGAN Inversion

Cited by 6 publications

References 32 publications

GaitEditer: Attribute Editing for Gait Representation Learning

GaitEditer: Attribute Editing for Gait Representation Learning

Gradual Residuals Alignment: A Dual-Stream Framework for GAN Inversion and Image Attribute Editing

Multi-domain Information Fusion for Key-Points Guided GAN Inversion

Contact Info

Product

Resources

About