Research map / Image generation / Editing and style transfer
FurryGAN: High Quality Foreground-Aware Image Synthesis
Jeongmin Bae, Mingi Kwon, Youngjung Uh
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 fine mask network (ablation) | FurryGAN · FFHQ ablation | FID: 9.48 -> 8.72 | Improved |
| add background participation loss (ablation) | FurryGAN · FFHQ ablation | FID: 9.79 -> 8.72 | Improved |
| add mask consistency loss (ablation) | FurryGAN · FFHQ ablation | FID: 9.53 -> 8.72 | Improved |
| add background participation loss (ablation) | FurryGAN · FFHQ ablation | mIoU: 0.84 -> 0.88 | Improved |
| add mask consistency loss (ablation) | FurryGAN · FFHQ ablation | mIoU: 0.86 -> 0.88 | Improved |
| add fine mask network (ablation) | FurryGAN · FFHQ ablation | mIoU: 0.88 -> 0.88 | No clear effect |
| add dual fake input (foreground and composite images both fed as fake) (ablation) | FurryGAN · FFHQ | mask separation | Improved |
| use FurryGAN instead of Labels4Free | FurryGAN · FFHQ (psi=1.0, TRACER pseudo-GT) | mIoU: 0.78 -> 0.88 | Improved |
| use FurryGAN instead of Labels4Free | FurryGAN · FFHQ | FID: 6.51 -> 8.72 | Hurt |
| use FurryGAN instead of Labels4Free | FurryGAN · AFHQv2-Cat | FID: 5.19 -> 6.34 | Hurt |
| use FurryGAN instead of Labels4Free | FurryGAN · CelebAMask-HQ inverted real images | mIoU: 0.87 -> 0.87 | No clear effect |
| use FurryGAN instead of PSeg | FurryGAN · FFHQ | FID: 62.44 -> 8.72 | Improved |
Closest papers by meaning
- Shadow Generation with Decomposed Mask Prediction and Attentive Shadow Filling (AAAI 2024 · similarity 0.84)
- MaskGAN: Towards Diverse and Interactive Facial Image Manipulation (CVPR 2020 · similarity 0.84)
- Intuitive, Interactive Beard and Hair Synthesis With Generative Models (CVPR 2020 · similarity 0.84)
- AnyScene: Customized Image Synthesis with Composited Foreground (CVPR 2024 · similarity 0.82)
- MonoHair: High-Fidelity Hair Modeling from a Monocular Video (CVPR 2024 · similarity 0.82)
- SmartMask: Context Aware High-Fidelity Mask Generation for Fine-grained Object Insertion and Layout Control (CVPR 2024 · similarity 0.82)
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.
