Research map / Image generation / Editing and style transfer
Be Your Own Prada: Fashion Synthesis With Structural Coherence
Shizhan Zhu, Raquel Urtasun, Sanja Fidler, Dahua Lin, Chen Change Loy
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 |
|---|---|---|---|
| use two-step shape-then-image generation instead of one-step GAN with 8x8x7 prior (ablation) | One-Step-8-7 · DeepFashion attribute detection on synthesized images | mAP (%): 72.4 -> 82.6 | Improved |
| use two-step shape-then-image generation instead of one-step GAN with merged 8x8x4 prior (ablation) | One-Step-8-4 · DeepFashion attribute detection on synthesized images | mAP (%): 73.3 -> 82.6 | Improved |
| add segmentation-guided compositional mapping (ablation) | Non-Compositional two-stage GAN · DeepFashion attribute detection on synthesized images | mAP (%): 74.7 -> 82.6 | Improved |
| add deep generative model over non-deep baseline | Non-Deep · DeepFashion attribute detection on synthesized images | mAP (%): 71.8 -> 82.6 | Improved |
| merge segmentation classes (8x8x4) instead of non-merged (8x8x7) prior in the one-step GAN (ablation) | One-Step · DeepFashion attribute detection on synthesized images | mAP (%): 72.4 -> 73.3 | No clear effect |
| add compositional mapping, T-Shirt attribute (ablation) | Non-Compositional · DeepFashion attribute detection on synthesized images (Has T-Shirt) | AP (%): 54.7 -> 63.2 | Improved |
| add compositional mapping, Has Long Pants (ablation) | Non-Compositional · DeepFashion attribute detection on synthesized images (Has Long Pants) | AP (%): 84.4 -> 90.7 | Improved |
| FashionGAN versus real image upper bound | FashionGAN · DeepFashion attribute detection on synthesized images | mAP (%): 86.8 -> 82.6 | Hurt |
| FashionGAN versus other methods in user ranking (2D non-parametric) | 2D Non-Parametric · User study ranking | mean rank: 2.286 -> 1.544 | Improved |
| add compositional mapping in user ranking (ablation) | Non-Compositional · User study ranking | mean rank: 3.045 -> 1.544 | Improved |
| generated segmentation maps vs real maps fool volunteers | FashionGAN shape generator · User study: segmentation maps | misclassification rate: 42% | Improved |
Closest papers by meaning
- A Generative Model of People in Clothing (ICCV 2017 · similarity 0.89)
- Weakly Supervised High-Fidelity Clothing Model Generation (CVPR 2022 · similarity 0.89)
- GarmentGPT: Compositional Garment Pattern Generation via Discrete Latent Tokenization (ICLR 2026 · similarity 0.88)
- Personalized Fashion Design (ICCV 2019 · similarity 0.88)
- Design2Cloth: 3D Cloth Generation from 2D Masks (CVPR 2024 · similarity 0.86)
- FashionR2R: Texture-preserving Rendered-to-Real Image Translation with Diffusion Models (NeurIPS 2024 · similarity 0.86)
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.
