Research map / Image generation / Editing and style transfer
UniCoRN: A Unified Conditional Image Repainting Network
Jimeng Sun, Shuchen Weng, Zheng Chang, Si Li, Boxin Shi
Check 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 |
|---|---|---|---|
| use UniCoRN instead of the strongest baseline | PERSON-CIR · PERSON-CIR | FID: 13.74 -> 11.45 | Improved |
| use UniCoRN instead of the strongest baseline | PERSON-CIR · PERSON-CIR | R-precision (%): 96.11 -> 97.42 | Improved |
| use UniCoRN instead of the strongest baseline | PERSON-CIR · PERSON-CIR | M-score: 3.86 -> 3.56 | Improved |
| use UniCoRN instead of the strongest baseline | LANDSCAPE-CIR · LANDSCAPE-CIR (256 × 256) | FID: 14.71 -> 9.96 | Improved |
| use UniCoRN instead of the strongest baseline | LANDSCAPE-CIR · LANDSCAPE-CIR (256 × 256) | R-precision (%): 96.73 -> 97.74 | Improved |
| use UniCoRN instead of the strongest baseline | LANDSCAPE-CIR · LANDSCAPE-CIR (256 × 256) | M-score: 3.18 -> 3.14 | Improved |
| use UniCoRN instead of the strongest baseline | LANDSCAPE-CIR · LANDSCAPE-CIR (512 × 512) | FID: 21.47 -> 18.63 | Improved |
| use UniCoRN instead of the strongest baseline | LANDSCAPE-CIR · LANDSCAPE-CIR (512 × 512) | R-precision (%): 96.96 -> 97.33 | Improved |
| use UniCoRN instead of the strongest baseline | LANDSCAPE-CIR · LANDSCAPE-CIR (512 × 512) | M-score: 15.66 -> 14.42 | Improved |
| use hierarchical HCMSM instead of CMSM (ablation) | UniCoRN · PERSON-CIR | FID: 15.03 -> 11.45 | Improved |
| use multiple encoder layers rather than a single layer in HCMSM (ablation) | UniCoRN · PERSON-CIR | FID: 12.40 -> 11.45 | Improved |
| connect the label-encoder units hierarchically (ablation) | UniCoRN · PERSON-CIR | FID: 12.21 -> 11.45 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SEAN — SEAN: Image Synthesis With Semantic Region-Adaptive Normalization (CVPR 2020) · better on 7 of 7 shared results
- MISC — MISC: Multi-Condition Injection and Spatially-Adaptive Compositing for Conditional Person Image Synthesis (CVPR 2020) · 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
- Prune and Repaint: Content-Aware Image Retargeting for any Ratio (NeurIPS 2024 · similarity 0.86)
- VIRES: Video Instance Repainting via Sketch and Text Guided Generation (CVPR 2025 · similarity 0.85)
- Training-and-Prompt-Free General Painterly Harmonization via Zero-Shot Disentenglement on Style and Content References (AAAI 2025 · similarity 0.84)
- ChromaFusionNet (CFNet): Natural Fusion of Fine-Grained Color Editing (AAAI 2024 · similarity 0.84)
- Training-free Dense-Aligned Diffusion Guidance for Modular Conditional Image Synthesis (CVPR 2025 · similarity 0.84)
- Spin: Diffusion-based Semantic Image Painting Through Independent Information Injection (AAAI 2025 · 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.
