Research map / Point clouds / Shape and surface completion
Style-Based Point Generator With Adversarial Rendering for Point Cloud Completion
Check 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 |
|---|---|---|---|
| add channel attention η to CAE (ablation) | SpareNet · ShapeNet | EMD × 10^3: 2.060 -> 1.862 | Helped |
| use EdgeConv edge features with channel attention instead of point features (ablation) | SpareNet · ShapeNet | EMD × 10^3: 1.972 -> 1.862 | Helped |
| use style-based folding instead of concatenation-based folding (ablation) | SpareNet · ShapeNet | EMD × 10^3: 3.274 -> 1.862 | Helped |
| use recurrent refinement instead of single-step refinement (ablation) | SpareNet · ShapeNet | EMD × 10^3: 2.184 -> 1.862 | Helped |
| apply adversarial rendering to the first refined output only, rather than no output (ablation) | SpareNet · ShapeNet | EMD × 10^3: 2.029 -> 1.862 | Helped |
| apply adversarial rendering to the final output only rather than the first refined output (ablation) | SpareNet · ShapeNet | EMD × 10^3: 1.862 -> 1.937 | Hurt |
| apply adversarial rendering to both refined outputs rather than the first output only (ablation) | SpareNet · ShapeNet | EMD × 10^3: 1.862 -> 1.898 | Hurt |
| increase adversarial loss weight from 0.1 to 0.2 (ablation) | SpareNet · ShapeNet | EMD × 10^3: 1.862 -> 1.898 | Hurt |
| increase adversarial loss weight from 0.1 to 5 (ablation) | SpareNet · ShapeNet | EMD × 10^3: 1.862 -> 1.980 | Hurt |
| use image-domain losses and a rendered-image discriminator instead of point-based supervision and discriminator (ablation) | SpareNet · ShapeNet | EMD × 10^3: 2.267 -> 1.862 | Helped |
| use image-domain losses and a rendered-image discriminator instead of point-based supervision and discriminator (ablation) | SpareNet · ShapeNet | FPD × 0.1: 2.120 -> 0.645 | Helped |
| use image-domain losses and a rendered-image discriminator instead of point-based supervision and discriminator (ablation) | SpareNet · ShapeNet | CD × 10^3: 0.706 -> 0.515 | Helped |
Closest papers by meaning
- Diverse Shape Completion via Style Modulated Generative Adversarial Networks (NeurIPS 2023 · similarity 0.91)
- Learning Progressive Point Embeddings for 3D Point Cloud Generation (CVPR 2021 · similarity 0.88)
- Orthogonal Dictionary Guided Shape Completion Network for Point Cloud (AAAI 2024 · similarity 0.86)
- DC-PCN: Point Cloud Completion Network with Dual-Codebook Guided Quantization (AAAI 2025 · similarity 0.86)
- CDPNet: Cross-Modal Dual Phases Network for Point Cloud Completion (AAAI 2024 · similarity 0.86)
- Hide in Thicket: Generating Imperceptible and Rational Adversarial Perturbations on 3D Point Clouds (CVPR 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.