Research map / Image generation / Editing and style transfer
Cross Attention Based Style Distribution for Controllable Person Image Synthesis
Xinyue Zhou, Mingyu Yin, Xinyuan Chen, Li Sun, Changxin Gao, Qingli Li
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 self-attention on style features in CASD block (ablation) | CASD · DeepFashion pose transfer | FID: 13.2462 -> 11.3732 | Improved |
| add self-attention on style features in CASD block (ablation) | CASD · DeepFashion pose transfer | SSIM: 0.7201 -> 0.7248 | Improved |
| add attention matrix prediction AM_p from target pose (ablation) | CASD · DeepFashion pose transfer | FID: 12.4265 -> 11.3732 | Improved |
| add attention matrix prediction AM_p from target pose (ablation) | CASD · DeepFashion pose transfer | LPIPS: 0.1985 -> 0.1936 | Improved |
| add AMCE loss guiding cross attention with parsing map (ablation) | CASD · DeepFashion pose transfer | FID: 14.7513 -> 11.3732 | Improved |
| add AMCE loss guiding cross attention with parsing map (ablation) | CASD · DeepFashion pose transfer | SSIM: 0.7156 -> 0.7248 | Improved |
| add AMCE loss guiding cross attention with parsing map (ablation) | CASD · DeepFashion pose transfer | PSNR: 31.41 -> 31.67 | Improved |
| use CASD instead of GFLA | pose transfer · DeepFashion pose transfer | FID: 10.5730 -> 11.3732 | Hurt |
| use CASD instead of SPGNet | pose transfer · DeepFashion pose transfer | user study Jab: 17.26% -> 28.96% | Improved |
| use CASD instead of PISE for virtual try-on (upper cloth) | try-on without retraining · Virtual try-on upper cloth | FID: 14.0537 -> 12.5376 | Improved |
| use CASD instead of SPGNet for parsing prediction | parsing map · DeepFashion parsing | average IoU: 61.89 -> 66.34 | 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.
- PISE — PISE: Person Image Synthesis and Editing With Decoupled GAN (CVPR 2021) · better on 6 of 6 shared results
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
- Coarse-to-Fine Latent Diffusion for Pose-Guided Person Image Synthesis (CVPR 2024 · similarity 0.86)
- Learning Flow Fields in Attention for Controllable Person Image Generation (CVPR 2025 · similarity 0.85)
- IMAGPose: A Unified Conditional Framework for Pose-Guided Person Generation (NeurIPS 2024 · similarity 0.85)
- HSI: A Holistic Style Injector for Arbitrary Style Transfer (CVPR 2025 · similarity 0.84)
- SCSA: A Plug-and-Play Semantic Continuous-Sparse Attention for Arbitrary Semantic Style Transfer (CVPR 2025 · similarity 0.84)
- Semantix: An Energy-guided Sampler for Semantic Style Transfer (ICLR 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.
