Research map / Point clouds / Shape and surface completion
Diverse Shape Completion via Style Modulated Generative Adversarial Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 style codes learned from complete shapes instead of Gaussian noise (ablation) | Diverse shape completion · PartNet | TMD: 3.71 -> 4.36 | Helped |
| Use style codes learned from complete shapes instead of a StyleGAN mapping network (ablation) | Diverse shape completion · PartNet | TMD: 4.03 -> 4.36 | Helped |
| Reduce style code dimension and add training noise (ablation) | Diverse shape completion · PartNet | TMD: 3.41 -> 4.36 | Helped |
| Replace the SeedFormer encoder with the proposed partial encoder (ablation) | SeedFormer-based completion network · PartNet | TMD: 3.21 -> 3.74 | Helped |
| Replace inverse-distance interpolation with PointConv interpolation (ablation) | SeedFormer-based completion network with partial encoder · PartNet | TMD: 3.74 -> 4.36 | Helped |
| Use discriminators and diversity penalties at multiple scales (ablation) | Diverse shape completion · PartNet | TMD: 2.05 -> 4.36 | Helped |
| Add completion loss L_comp (ablation) | Diverse shape completion · PartNet | MMD: 1.62 -> 1.50 | Helped |
| Add partial reconstruction loss L_part (ablation) | Diverse shape completion · PartNet | UHD: 13.03 -> 3.79 | Helped |
| Add diversity loss L_div (ablation) | Diverse shape completion · PartNet | TMD: 0.41 -> 4.36 | Helped |
| Compute diversity penalty in discriminator feature space rather than output-space EMD (ablation) | Diverse shape completion · PartNet | MMD: 1.82 -> 1.50 | Helped |
| Compute diversity penalty in discriminator feature space rather than output-space EMD (ablation) | Diverse shape completion · PartNet | UHD: 6.16 -> 3.79 | Helped |
| Use output-space EMD diversity penalty instead of discriminator feature-space penalty (ablation) | Diverse shape completion · PartNet | TMD: 4.36 -> 7.14 | Mixed |
Closest papers by meaning
- Style-Based Point Generator With Adversarial Rendering for Point Cloud Completion (CVPR 2021 · similarity 0.91)
- Unsupervised 3D Shape Completion Through GAN Inversion (CVPR 2021 · similarity 0.91)
- Large Scale Image Completion via Co-Modulated Generative Adversarial Networks (ICLR 2021 · similarity 0.89)
- CDPNet: Cross-Modal Dual Phases Network for Point Cloud Completion (AAAI 2024 · similarity 0.88)
- PCDreamer: Point Cloud Completion Through Multi-view Diffusion Priors (CVPR 2025 · similarity 0.87)
- Learning Generalizable Shape Completion with SIM(3) Equivariance (NeurIPS 2025 · similarity 0.87)
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.