Research map / Image generation / GANs
IntereStyle: Encoding an Interest Region for Robust StyleGAN Inversion
Seung-Jun Moon, Gyeong-Moon Park
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add L_image on the interest region (ablation) | Restyle · CelebA-HQ interest region | L2: 0.015 -> 0.013 | Improved |
| add L_image on the interest region (ablation) | Restyle · CelebA-HQ face region | L2: 0.004 -> 0.005 | Hurt |
| add L_image on the interest region (ablation) | Restyle · CelebA-HQ interest region | LPIPS: 0.088 -> 0.084 | Improved |
| add Interest Disentanglement after L_image (ablation) | Restyle · CelebA-HQ interest region | L2: 0.013 -> 0.012 | Improved |
| add Interest Disentanglement after L_image (ablation) | Restyle · CelebA-HQ face region | L2: 0.005 -> 0.003 | Improved |
| add Interest Disentanglement after L_image (ablation) | Restyle · CelebA-HQ interest region | LPIPS: 0.084 -> 0.078 | Improved |
| add Uninterest Filter after InD (ablation) | Restyle · CelebA-HQ interest region | LPIPS: 0.078 -> 0.075 | Improved |
| add Uninterest Filter after InD (ablation) | Restyle · CelebA-HQ face region | LPIPS: 0.037 -> 0.036 | Improved |
| add Uninterest Filter after InD (ablation) | Restyle · CelebA-HQ interest region | L2: 0.012 -> 0.013 | Hurt |
| use interest-region-focused inversion (IntereStyle) instead of Restyle_pSp | IntereStyle · CelebA-HQ ID similarity | ID similarity: 0.65 -> 0.68 | Improved |
| use mask dilation on the interest region (ablation) | IntereStyle · CelebA-HQ face boundary | boundary reconstruction error (qualitative) | Improved |
Closest papers by meaning
- ReStyle: A Residual-Based StyleGAN Encoder via Iterative Refinement (ICCV 2021 · similarity 0.88)
- In-Domain GAN Inversion for Faithful Reconstruction and Editability (TPAMI 2024 · similarity 0.88)
- Delving StyleGAN Inversion for Image Editing: A Foundation Latent Space Viewpoint (CVPR 2023 · similarity 0.88)
- Diverse Inpainting and Editing with GAN Inversion (ICCV 2023 · similarity 0.88)
- In-Domain GAN Inversion for Real Image Editing (ECCV 2020 · similarity 0.87)
- In-N-Out: Faithful 3D GAN Inversion with Volumetric Decomposition for Face Editing (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
