Research map / Image generation / Autoregressive and latent generation
Retrieval-Based Spatially Adaptive Normalization for Semantic Image Synthesis
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use RESAIL pixel-level normalization with retrieval guidance instead of SPADE normalization (ablation) | Semantic image synthesis on Cityscapes · Cityscapes | FID: 58.7 -> 45.5 | Helped |
| Inject retrieval guidance through RESAIL rather than concatenating it into a convolution layer (ablation) | Semantic image synthesis on Cityscapes · Cityscapes | FID: 47.8 -> 45.5 | Helped |
| Inject retrieval guidance through RESAIL rather than SPADE modulation (ablation) | Semantic image synthesis on Cityscapes · Cityscapes | FID: 53.4 -> 45.5 | Helped |
| Inject retrieval guidance through RESAIL rather than SEAN modulation (ablation) | Semantic image synthesis on Cityscapes · Cityscapes | FID: 66.6 -> 45.5 | Helped |
| Add distorted ground-truth guidance alongside retrieval guidance during training (ablation) | RESAIL on Cityscapes · Cityscapes | FID: 47.7 -> 45.5 | Helped |
| Use distorted rather than original ground-truth guidance alongside retrieval guidance (ablation) | RESAIL on Cityscapes · Cityscapes | FID: 52.8 -> 45.5 | Helped |
| Add segmentation loss (ablation) | RESAIL on Cityscapes · Cityscapes | FID: 46.8 -> 45.5 | Helped |
| Raise the retrieval non-similarity threshold, allowing more dissimilar guidance (ablation) | RESAIL on Cityscapes · Cityscapes | FID: 45.49 -> 51.04 | Hurt |
| Use RESAIL instead of OASIS | Semantic image synthesis on ADE20K-outdoor · ADE20K-outdoor | AC: 86.5 | Helped |
| Use RESAIL instead of OASIS | Semantic image synthesis on Cityscapes · Cityscapes | FID: 47.7 -> 45.5 | Helped |
| Use RESAIL instead of OASIS | Semantic image synthesis on COCO-Stuff · COCO-Stuff | mIOU: 44.1 -> 44.7 | Helped |
Closest papers by meaning
- PLACE: Adaptive Layout-Semantic Fusion for Semantic Image Synthesis (CVPR 2024 · similarity 0.86)
- TexTailor: Customized Text-aligned Texturing via Effective Resampling (ICLR 2025 · similarity 0.84)
- UniFlow: A Unified Pixel Flow Tokenizer for Visual Understanding and Generation (ICLR 2026 · similarity 0.84)
- SeD: Semantic-Aware Discriminator for Image Super-Resolution (CVPR 2024 · similarity 0.84)
- STARFlow: Scaling Latent Normalizing Flows for High-resolution Image Synthesis (NeurIPS 2025 · similarity 0.84)
- SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval (ICML 2025 · similarity 0.83)
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.