Research map / Image generation / Editing and style transfer
Deep Image Spatial Transformation for Person Image Generation
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 global-flow local-attention instead of the strongest reported baseline | Pose-guided person image generation · DeepFashion | FID: 18.457 -> 10.573 | Helped |
| Use global-flow local-attention instead of the strongest reported baseline | Pose-guided person image generation · DeepFashion | LPIPS: 0.2131 -> 0.2341 | Hurt |
| Use global-flow local-attention instead of the strongest reported baseline | Pose-guided person image generation · DeepFashion | JND: 12.61% -> 24.80% | Helped |
| Use global-flow local-attention instead of the strongest reported baseline | Pose-guided person image generation · Market-1501 | FID: 20.144 -> 19.751 | Helped |
| Use global-flow local-attention instead of the strongest reported baseline | Pose-guided person image generation · Market-1501 | LPIPS: 0.2888 -> 0.2817 | Helped |
| Use global-flow local-attention instead of the strongest reported baseline | Pose-guided person image generation · Market-1501 | Mask-LPIPS: 0.1403 -> 0.1482 | Hurt |
| Use global-flow local-attention instead of the strongest reported baseline | Pose-guided person image generation · Market-1501 | JND: 30.85% -> 27.81% | Hurt |
| Add flow-based spatial transformation to the convolutional baseline (ablation) | Person image generation on DeepFashion · DeepFashion | FID: 16.008 -> 12.143 | Helped |
| Add flow-based spatial transformation to the convolutional baseline (ablation) | Person image generation on DeepFashion · DeepFashion | LPIPS: 0.2473 -> 0.2406 | Helped |
| Replace local attention with global attention (ablation) | Full method on DeepFashion · DeepFashion | FID: 10.573 -> 18.616 | Hurt |
| Replace local attention with global attention (ablation) | Full method on DeepFashion · DeepFashion | LPIPS: 0.2341 -> 0.2575 | Hurt |
| Add content-aware sampling instead of bilinear sampling (ablation) | Flow-based method on DeepFashion · DeepFashion | FID: 12.143 -> 10.573 | Helped |
Closest papers by meaning
- Structure-aware Person Image Generation with Pose Decomposition and Semantic Correlation (AAAI 2021 · similarity 0.89)
- Learning Flow Fields in Attention for Controllable Person Image Generation (CVPR 2025 · similarity 0.87)
- IMAGPose: A Unified Conditional Framework for Pose-Guided Person Generation (NeurIPS 2024 · similarity 0.86)
- Exploring Dual-Task Correlation for Pose Guided Person Image Generation (CVPR 2022 · similarity 0.86)
- Coarse-to-Fine Latent Diffusion for Pose-Guided Person Image Synthesis (CVPR 2024 · similarity 0.85)
- Advancing Pose-Guided Image Synthesis with Progressive Conditional Diffusion Models (ICLR 2024 · similarity 0.84)
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.