Research map / Image generation / Editing and style transfer
Particularity beyond Commonality: Unpaired Identity Transfer with Multiple References
Ruizheng Wu, Xin Tao, Yingcong Chen, Xiaoyong Shen, Jiaya Jia
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 category classification discriminator Dcl (ablation) | multi-reference framework · RaFD | Acc (%): 16.77 -> 73.03 | Improved |
| add category classification discriminator Dcl (ablation) | multi-reference framework · RaFD | IS: 5.80 -> 2.49 | Hurt |
| add category classification discriminator Dcl (ablation) | multi-reference framework · RaFD | mFID (1e3): 12.13 -> 6.08 | Improved |
| add category comparison discriminator Dco (ablation) | multi-reference framework · RaFD | Acc (%): 68.84 -> 73.03 | Improved |
| add category comparison discriminator Dco (ablation) | multi-reference framework · RaFD | mFID (1e3): 7.70 -> 6.08 | Improved |
| add category comparison discriminator Dco (ablation) | multi-reference framework · RaFD | IS: 2.41 -> 2.49 | Improved |
| increase reference images from 1 to 3 (ablation) | multi-reference framework · RaFD | Acc (%): 53.12 -> 73.03 | Improved |
| increase reference images from 3 to 5 (ablation) | multi-reference framework · RaFD | Acc (%): 73.03 -> 90.01 | Improved |
| increase reference images from 5 to 10 (ablation) | multi-reference framework · RaFD | Acc (%): 90.01 -> 90.21 | No clear effect |
| increase reference images from 1 to 10 (ablation) | multi-reference framework · RaFD | Runtime (ms): 51.1 -> 71.5 | Hurt |
| increase reference images from 1 to 5 (ablation) | multi-reference framework · RaFD | IS: 3.31 -> 1.46 | Hurt |
| use semantic pyramid alignment instead of only highest-level features (ablation) | multi-reference framework · RaFD (qualitative) | visual quality | Improved |
Closest papers by meaning
- WithAnyone: Toward Controllable and ID Consistent Image Generation (ICLR 2026 · similarity 0.86)
- Infinite-ID: Identity-preserved Personalization via ID-semantics Decoupling Paradigm (ECCV 2024 · similarity 0.84)
- Semantically Robust Unpaired Image Translation for Data With Unmatched Semantics Statistics (ICCV 2021 · similarity 0.84)
- TopoRefine: Topology-Aware Correspondence and Residual Refinement for Training-Free Subject-Consistent Generation (NeurIPS 2026 · similarity 0.84)
- Exploring Unlabeled Faces for Novel Attribute Discovery (CVPR 2020 · similarity 0.84)
- ID-Patch: Robust ID Association for Group Photo Personalization (CVPR 2025 · similarity 0.83)
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.
