Research map / Image generation / Editing and style transfer
Pose-Star: Anatomy-Aware Editing for Open-World Fashion Images
Yuran Dong, Mang Ye
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 |
|---|---|---|---|
| set star-region radius to the average instead of minimum (ablation) | Pose-Star · Keypoint radius ablation | average IoU: 0.69 -> 0.73 | Improved |
| set star-region radius to the average instead of maximum (ablation) | Pose-Star · Keypoint radius ablation | average IoU: 0.67 -> 0.73 | Improved |
| use 3x3 sliding window instead of 1x1 for multi-stage attention fusion (ablation) | Pose-Star · Sliding window size ablation | average IoU: 0.51 -> 0.73 | Improved |
| use 3x3 sliding window instead of 2x2 (ablation) | Pose-Star · Sliding window size ablation | average IoU: 0.67 -> 0.73 | Improved |
| enlarge sliding window from 3x3 to 4x4 (ablation) | Pose-Star · Sliding window size ablation | average IoU: 0.73 -> 0.69 | Hurt |
| replace Pose-Star aggregation with SCHP fixed parsing (ablation) | Pose-Star · Aggregation vs segmenters | valid region retention (%): 92.5% -> 48.2% | Hurt |
| replace Pose-Star aggregation with SAM (ablation) | Pose-Star · Aggregation vs segmenters | boundary jitter: 62% less boundary jitter than SAM | Hurt |
| replace Pose-Star edge optimization with DIS/U2-Net (ablation) | Pose-Star · Optimization vs edge detectors | hand/foot distortion: reduces hand/foot distortions by 35% compared to DIS/U2-Net | Hurt |
| add Pose-Star fine-grained human masks over fixed-mask editors | virtual try-on and text editors · Wild-image test set (16,800 samples) | ImageReward: 1.35 -> 1.61 | Improved |
| plug Pose-Star into existing editors | IDM-VTON and PowerPaint · User study (200 participants) | user rating (0-5): +1.81 points pose robustness, +0.81 in-the-wild, +3.11 user-defined flexibility | 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.
- TexFit — TexFit: Text-Driven Fashion Image Editing with Diffusion Models (AAAI 2024) · better on 1 of 1 shared result
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
- FashionTailor: Controllable Clothing Editing for Human Images with Appearance Preserving (AAAI 2025 · similarity 0.90)
- TexFit: Text-Driven Fashion Image Editing with Diffusion Models (AAAI 2024 · similarity 0.87)
- UniHuman: A Unified Model For Editing Human Images in the Wild (CVPR 2024 · similarity 0.87)
- Disentangled Pose and Appearance Guidance for Multi-Pose Generation (CVPR 2025 · similarity 0.87)
- IMAGDressing-v1: Customizable Virtual Dressing (AAAI 2025 · similarity 0.86)
- ChatGarment: Garment Estimation, Generation and Editing via Large Language Models (CVPR 2025 · similarity 0.85)
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.
