Research map / Image generation / Editing and style transfer
Semantic Layout Manipulation With High-Resolution Sparse Attention
Haitian Zheng, Zhe Lin, Jingwan Lu, Scott D. Cohen, Jianming Zhang, Ning Xu, Jiebo Luo
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 high-resolution sparse-attention correspondence (vs none) (ablation) | Ours · Places365 manipulation | FID: 38.61 -> 31.35 | Improved |
| add high-resolution sparse-attention correspondence (vs none) (ablation) | Ours · Places365 manipulation | PSNR: 22.999 -> 24.252 | Improved |
| use high-resolution instead of low-resolution CoCosNet correspondence (ablation) | Ours · Places365 manipulation | FID: 32.65 -> 31.35 | Improved |
| use high-resolution instead of low-resolution CoCosNet correspondence (ablation) | Ours · Places365 manipulation | PSNR: 23.751 -> 24.252 | Improved |
| add semantic encoder branch (ablation) | Ours · Places365 manipulation | FID: 33.51 -> 31.35 | Improved |
| add contextual attention branch (ablation) | Ours · Places365 manipulation | FID: 32.18 -> 31.35 | Improved |
| add dilated convolution branch (ablation) | Ours · Places365 manipulation | FID: 32.43 -> 31.35 | Improved |
| feed intermediate coarse-stage features to refinement network (ablation) | Ours · Places365 manipulation | FID: 33.02 -> 31.35 | Improved |
| add semantic encoder branch (ablation) | Ours · Places365 manipulation | style loss: 4.875e-07 -> 6.801e-07 | Hurt |
| use sparse-attention warping method vs CoCosNet | Ours · ADE20k manipulation | FID: 98.13 -> 28.07 | Improved |
| use sparse-attention method vs CoCosNet | Ours · ADE20k manipulation | PSNR: 24.063 -> 22.672 | Hurt |
| use sparse attention (0.4 s) vs optimization-based deep image analogy (10.5 minutes) for correspondence | Ours · correspondence computation | time: 10.5 minutes -> 0.4 seconds | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SESAME — SESAME: Semantic Editing of Scenes by Adding, Manipulating or Erasing Objects (ECCV 2020) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PLACE: Adaptive Layout-Semantic Fusion for Semantic Image Synthesis (CVPR 2024 · similarity 0.89)
- SSMG: Spatial-Semantic Map Guided Diffusion Model for Free-Form Layout-to-Image Generation (AAAI 2024 · similarity 0.88)
- Envisioning Beyond the Few: Disentangled Semantics and Primitives for Few-Shot Atypical Layout-to-Image Generation (ICML 2026 · similarity 0.88)
- Rethinking The Training And Evaluation of Rich-Context Layout-to-Image Generation (NeurIPS 2024 · similarity 0.87)
- Control and Realism: Best of Both Worlds in Layout-to-Image without Training (ICML 2025 · similarity 0.87)
- SpotActor: Training-Free Layout-Controlled Consistent Image Generation (AAAI 2025 · 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.
