Research map / 3D vision / Face and shape
Multimodal Shape Completion via Conditional Generative Adversarial Networks
Rundi Wu, Xuelin Chen, Yixin Zhuang, Baoquan Chen
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 explicit mode encoder (shape VAE encoder) instead of implicit point-cloud mode encoder (ablation) | Ours-im-pc2z · PartNet | TMD x10^2 (diversity): 0.69 -> 3.12 | Improved |
| add explicit mode encoder instead of implicit point-cloud mode encoder (ablation) | Ours-im-pc2z · PartNet | MMD x10^3 (quality): 2.00 -> 1.65 | Improved |
| add explicit mode encoder instead of implicit point-cloud mode encoder (ablation) | Ours-im-pc2z · PartNet-Scan | TMD x10^2 (diversity): 1.42 -> 3.65 | Improved |
| add explicit mode encoder instead of implicit point-cloud mode encoder (ablation) | Ours-im-pc2z · 3D-EPN | TMD x10^2 (diversity): 1.33 -> 3.03 | Improved |
| add explicit mode encoder instead of implicit latent-code mode encoder (ablation) | Ours-im-l2z · PartNet | UHD x10^2 (fidelity, lower better): 7.33 -> 6.06 | Improved |
| add explicit mode encoder instead of implicit latent-code mode encoder (ablation) | Ours-im-l2z · PartNet | MMD x10^3 (quality): 1.93 -> 1.65 | Improved |
| add explicit mode encoder instead of implicit latent-code mode encoder (ablation) | Ours-im-l2z · PartNet | TMD x10^2 (diversity): 3.34 -> 3.12 | Hurt |
| add explicit mode encoder instead of implicit latent-code mode encoder (ablation) | Ours-im-l2z · 3D-EPN | MMD x10^3 (quality): 1.80 -> 1.67 | Improved |
| use learned conditional GAN instead of nearest-neighbor retrieval in latent space | KNN-latent · PartNet | MMD x10^3 (quality): 1.47 -> 1.65 | Hurt |
| increase weight beta of latent regression loss (ablation) | Ours · PartNet | TMD (diversity) and UHD (fidelity) | Mixed |
| increase weight of reconstruction loss (ablation) | Ours · PartNet | diversity of completions | Hurt |
| increase number of missing parts in input | Ours · PartNet | TMD (diversity) | Improved |
Closest papers by meaning
- Diverse Shape Completion via Style Modulated Generative Adversarial Networks (NeurIPS 2023 · similarity 0.92)
- Learning Shape Priors for Single-View 3D Completion and Reconstruction (ECCV 2018 · similarity 0.91)
- DiffComplete: Diffusion-based Generative 3D Shape Completion (NeurIPS 2023 · similarity 0.90)
- Unsupervised 3D Shape Completion Through GAN Inversion (CVPR 2021 · similarity 0.89)
- Denoise and Contrast for Category Agnostic Shape Completion (CVPR 2021 · similarity 0.89)
- CDPNet: Cross-Modal Dual Phases Network for Point Cloud Completion (AAAI 2024 · 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.
