Real image editing with StyleGAN
| buir.advisor | Boral, Ayşegül Dündar | |
| dc.contributor.author | Pehlivan, Hamza | |
| dc.date.accessioned | 2023-09-13T11:16:15Z | |
| dc.date.available | 2023-09-13T11:16:15Z | |
| dc.date.copyright | 2023-09 | |
| dc.date.issued | 2023-09 | |
| dc.date.submitted | 2023-09-11 | |
| dc.department | Department of Computer Engineering | |
| dc.description | Cataloged from PDF version of article. | |
| dc.description | Thesis (Master's): Bilkent University, Department of Computer Engineering, İhsan Doğramacı Bilkent University, 2023. | |
| dc.description | Includes bibliographical references (leaves 37-42). | |
| dc.description.abstract | We 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 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. | |
| dc.description.degree | M.S. | |
| dc.description.statementofresponsibility | by Hamza Pehlivan | |
| dc.format.extent | x, 42 leaves : charts, photography ; 30 cm. | |
| dc.identifier.itemid | B162502 | |
| dc.identifier.uri | https://hdl.handle.net/11693/113861 | |
| dc.language.iso | English | |
| dc.publisher | Bilkent University | |
| dc.rights | info:eu-repo/semantics/openAccess | |
| dc.subject | Generative Adversarial Networks (GANs) | |
| dc.subject | GAN Inversion | |
| dc.subject | Image editing with GANs | |
| dc.title | Real image editing with StyleGAN | |
| dc.title.alternative | StyleGAN ile gerçek resim düzenleme | |
| dc.type | Thesis |