Research map / 3D vision / Face and shape
A Generative Model of People in Clothing
Christoph Lassner, Gerard Pons-Moll, Peter V. Gehler
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 |
|---|---|---|---|
| train segmentation on real instead of fully synthetic images, test on real | DeepLab ResNet-101 · Segmentation test on real | IoU: 0.335 -> 0.522 | Improved |
| train segmentation on fully synthetic data, test on fully synthetic | DeepLab ResNet-101 · Segmentation test on full synthetic | accuracy: 0.955 -> 0.978 | No clear effect |
| train on fully synthetic data instead of real, test on real | DeepLab ResNet-101 · Segmentation test on real | accuracy: 0.951 -> 0.898 | Hurt |
| train on synthetic textures (portray on GT sketches) instead of real, test on real | DeepLab ResNet-101 · Segmentation test on real | accuracy: 0.951 -> 0.915 | Hurt |
| replace ground-truth sketches by generated sketches in the portray module input (ablation) | ClothNet · User study | generated image rated real: 0.413 -> 0.247 | Hurt |
| condition the sketch VAE on a six-part SMPL silhouette (CSM vs LSM) | sketch module · Chictopia10K test reconstruction | F1: 0.510 -> 0.488 | Hurt |
| use two-stage sketch VAE plus portray instead of direct image-to-image translation from SMPL sketches (ablation) | ClothNet · qualitative | image quality | Improved |
| add skip connections (image-to-image network) instead of context encoder (ablation) | portray module · qualitative | level of detail | Improved |
| use six-part body silhouette instead of foreground-background or finer segmentation (ablation) | conditional sketch module · qualitative | conditioning quality | Improved |
| add facial landmark annotations from face shape matcher (ablation) | sketch and portray modules · qualitative | face realism | Improved |
Closest papers by meaning
- Learning to Dress 3D People in Generative Clothing (CVPR 2020 · similarity 0.91)
- SiCloPe: Silhouette-Based Clothed People (CVPR 2019 · similarity 0.91)
- 3DPeople: Modeling the Geometry of Dressed Humans (ICCV 2019 · similarity 0.91)
- SMPLicit: Topology-Aware Generative Model for Clothed People (CVPR 2021 · similarity 0.91)
- FreeCloth: Free-form Generation Enhances Challenging Clothed Human Modeling (CVPR 2025 · similarity 0.90)
- Controllable Human Image Generation with Personalized Multi-Garments (CVPR 2025 · similarity 0.88)
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.
