Research map / Detection and segmentation / CNN features
FashionMirror: Co-Attention Feature-Remapping Virtual Try-On With Sequential Template Poses
Chieh-Yun Chen, Ling Lo, Pin-Jui Huang, Hong-Han Shuai, Wen-Huang Cheng
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 per-frame source human feature enhancement (hs boost) (ablation) | FashionMirror · Fashion video try-on test set (2000 clips) | LPIPS: 0.073 -> 0.057 | Improved |
| add per-frame source human feature enhancement (hs boost) (ablation) | FashionMirror · Fashion video try-on test set (2000 clips) | VFID (I3D): 4.593 -> 3.097 | Improved |
| add per-frame source human feature enhancement (hs boost) (ablation) | FashionMirror · Fashion video try-on test set (2000 clips) | SSIM: 0.906 -> 0.923 | Improved |
| add layer weighting lambda_ci in clothing feature fusion (ablation) | FashionMirror · Fashion video try-on test set (2000 clips) | LPIPS: 0.060 -> 0.057 | Improved |
| add layer weighting lambda_ci in clothing feature fusion (ablation) | FashionMirror · Fashion video try-on test set (2000 clips) | VFID (3D-ResNet): 1.206 -> 1.033 | Improved |
| add layer weighting lambda_ci in clothing feature fusion (ablation) | FashionMirror · Fashion video try-on test set (2000 clips) | IS: 2.187 -> 2.234 | Hurt |
| use single-flow STN instead of multi-flow feature enhancement (ablation) | FashionMirror · Fashion video try-on test set (2000 clips) | VFID (I3D): 5.226 -> 3.097 | Improved |
| use single-flow STN instead of multi-flow feature enhancement (ablation) | FashionMirror · Fashion video try-on test set (2000 clips) | LPIPS: 0.092 -> 0.057 | Improved |
| use co-attention mask network instead of semantic segmentation | try-on preprocessing · Runtime (40,000 input sets, 2080-ti) | seconds per input: 0.3469 -> 0.1983 | Improved |
| fuse at feature level with FashionMirror instead of pixel-level pipelines (vs ACGPN + GFLA) | FashionMirror · Fashion video try-on test set (2000 clips) | VFID (I3D): 4.809 -> 3.097 | Improved |
| FashionMirror versus FashionOn on Inception Score | FashionMirror · Fashion video try-on test set (2000 clips) | IS: 2.388 -> 2.234 | Hurt |
Closest papers by meaning
- Toward Characteristic-Preserving Image-based Virtual Try-On Network (ECCV 2018 · similarity 0.89)
- Towards Photo-Realistic Virtual Try-On by Adaptively Generating-Preserving Image Content (CVPR 2020 · similarity 0.89)
- Full-Range Virtual Try-On With Recurrent Tri-Level Transform (CVPR 2022 · similarity 0.88)
- Robust-MVTON: Learning Cross-Pose Feature Alignment and Fusion for Robust Multi-View Virtual Try-On (CVPR 2025 · similarity 0.87)
- MV-VTON: Multi-View Virtual Try-On with Diffusion Models (AAAI 2025 · similarity 0.87)
- IMAGDressing-v1: Customizable Virtual Dressing (AAAI 2025 · similarity 0.87)
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.
