Research map / 3D vision / Human pose estimation
Collecting The Puzzle Pieces: Disentangled Self-Driven Human Pose Transfer by Permuting Textures
Nannan Li, Kevin J Shih, Bryan A. Plummer
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 |
|---|---|---|---|
| disentangle pose from texture by permuting image patches instead of part-wise encoding (ablation) | PT2 · DeepFashion | FID: 10.279 -> 8.338 | Improved |
| use patch permutation instead of TPS input warping (ablation) | PT2 · DeepFashion | FID: 10.011 -> 8.338 | Improved |
| add TPS warping on top of input permutation (ablation) | PT2 · DeepFashion | FID: 8.338 -> 8.332 | No clear effect |
| add source pose masking (50%) (ablation) | PT2 · DeepFashion | FID: 9.451 -> 8.338 | Improved |
| add texture masking (20%) (ablation) | PT2 · DeepFashion | FID: 8.780 -> 8.338 | Improved |
| add parsing map input (ablation) | PT2 · DeepFashion | FID: 8.460 -> 8.338 FID; IS 3.302 -> 3.469 | Mixed |
| add small-kernel encoder to dual-scale encoder (ablation) | PT2 · DeepFashion | FID: 8.905 -> 8.338 | Improved |
| add large-kernel encoder to dual-scale encoder (ablation) | PT2 · DeepFashion | FID: 9.275 -> 8.338 | Improved |
| use dual-scale encoder instead of encoding each patch separately (ablation) | PT2 · DeepFashion | FID: 8.514 -> 8.338 | Improved |
| use separate source pose branch instead of concatenating pose with texture (ablation) | PT2 · DeepFashion | FID: 9.236 -> 8.338 | Improved |
| add texture-to-pose matching in cross-attention (ablation) | PT2 · DeepFashion | FID: 10.786 -> 8.338 | Improved |
| use patch size 16 rather than 8 or 64 (ablation) | PT2 · DeepFashion | FID: 8.565 (p=8) / 8.898 (p=64) -> 8.338 (p=16) | Improved |
Closest papers by meaning
- SwapNet: Garment Transfer in Single View Images (ECCV 2018 · similarity 0.88)
- Pose-Guided Human Animation From a Single Image in the Wild (CVPR 2021 · similarity 0.88)
- Coordinate-Based Texture Inpainting for Pose-Guided Human Image Generation (CVPR 2019 · similarity 0.88)
- D^3-Human: Dynamic Disentangled Digital Human from Monocular Video (CVPR 2025 · similarity 0.86)
- VS: Reconstructing Clothed 3D Human from Single Image via Vertex Shift (CVPR 2024 · similarity 0.86)
- Disentangled Pose and Appearance Guidance for Multi-Pose Generation (CVPR 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.
