StyleRes: transforming the residuals for real ımage editing with StyleGAN

buir.contributor.authorPehlivan, Hamza
buir.contributor.authorDalva, Yusuf
buir.contributor.authorDündar, Aysegül
buir.contributor.orcidDalva, Yusuf|0000-0002-8402-8291
buir.contributor.orcidDündar, Aysegül|0000-0003-2014-6325
dc.citation.epage1837en_US
dc.citation.spage1828
dc.contributor.authorPehlivan, Hamza
dc.contributor.authorDalva, Yusuf
dc.contributor.authorDündar, Aysegül
dc.coverage.spatialVancouver, BC, Canada
dc.date.accessioned2024-03-07T08:40:08Z
dc.date.available2024-03-07T08:40:08Z
dc.date.issued2023-07-22
dc.departmentDepartment of Computer Engineering
dc.descriptionConference Name: 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
dc.descriptionDate of Conference: 17-24 June 2023
dc.description.abstractWe present a novel image inversion framework and a training pipeline to achieve high-fidelity image inversion with high-quality attribute editing. Inverting real images into StyleGAN’s latent space is an extensively studied problem, yet the trade-off between the image reconstruction fidelity and image editing quality remains an open challenge. The low-rate latent spaces are limited in their expressiveness power for high-fidelity reconstruction. On the other hand, high-rate latent spaces result in degradation in editing quality. In this work, to achieve high-fidelity inversion, we learn residual features in higher latent codes that lower latent codes were not able to encode. This enables preserving image details in reconstruction. To achieve high-quality editing, we learn how to transform the residual features for adapting to manipulations in latent codes. We train the framework to extract residual features and transform them via a novel architecture pipeline and cycle consistency losses. We run extensive experiments and compare our method with state-of-the-art inversion methods. Qualitative metrics and visual comparisons show significant improvements. Code: https://github.com/hamzapehlivan/StyleRes
dc.description.provenanceMade available in DSpace on 2024-03-07T08:40:08Z (GMT). No. of bitstreams: 1 StyleRes_transforming_the_residuals_for_real_ımage_editing_with_StyleGAN.pdf: 3077526 bytes, checksum: b2f6914c768d82800b9eb30dc98e0d3b (MD5) Previous issue date: 2023-07-22en
dc.identifier.doi10.1109/CVPR52729.2023.00182en_US
dc.identifier.eisbn979-8-3503-0129-8en_US
dc.identifier.isbn979-8-3503-0130-4en_US
dc.identifier.issn1063-6919en_US
dc.identifier.issn2575-7075en_US
dc.identifier.urihttps://hdl.handle.net/11693/114379en_US
dc.language.isoEnglishen_US
dc.publisherIEEEen_US
dc.relation.isversionofhttps://dx.doi.org/10.1109/CVPR52729.2023.00182
dc.source.title2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
dc.subjectTraining
dc.subjectMeasurement
dc.subjectDegradation
dc.subjectVisualization
dc.subjectCodes
dc.subjectPipelines
dc.subjectTransforms
dc.titleStyleRes: transforming the residuals for real ımage editing with StyleGAN
dc.typeConference Paper

Files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
StyleRes_transforming_the_residuals_for_real_ımage_editing_with_StyleGAN.pdf
Size:
2.93 MB
Format:
Adobe Portable Document Format

License bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
2.01 KB
Format:
Item-specific license agreed upon to submission
Description: