Research map / Point clouds / Shape and surface completion
CAPNet: Continuous Approximation Projection for 3D Point Cloud Reconstruction Using 2D Supervision
K. L. Navaneet, Priyanka Mandikal, Mayank Agarwal, R. Venkatesh Babu
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 CAPNet with mask supervision instead of DRC-Mask | ShapeNet reconstruction · ShapeNet (mean) | Chamfer distance (×1000): 13.26 -> 2.69 | Improved |
| Use CAPNet with mask supervision instead of Lin et al. | ShapeNet reconstruction · ShapeNet (mean) | Chamfer distance (×1000): 3.62 -> 2.69 | Improved |
| Fine-tune PSGN-FC using 2D mask supervision | PSGN-FC · Pix3D test | Chamfer distance (×1000): 10.1 -> 9.2 | Improved |
| Optimize the encoder while fixing the decoder using the test mask (ablation) | Pre-trained PSGN-FC · Test-stage optimization | Chamfer distance (×1000): 4.64 -> 4.43 | Improved |
| Optimize both encoder and decoder with regularization using the test mask | Pre-trained PSGN-FC · Test-stage optimization | Chamfer distance (×1000): 4.64 -> 4.37 | Improved |
| Optimize predicted points directly using the test mask (ablation) | Pre-trained PSGN-FC · Test-stage optimization | Chamfer distance (×1000): 4.64 -> 4.49 | Improved |
| Increase mask supervision from one view to two views (ablation) | CAPNet · ShapeNet airplane | Chamfer distance (×1000): 2.40 -> 2.02 | Improved |
| Increase mask supervision from one view to two views (ablation) | CAPNet · ShapeNet car | Chamfer distance (×1000): 3.47 -> 1.68 | Improved |
| Increase mask supervision from one view to two views (ablation) | CAPNet · ShapeNet chair | Chamfer distance (×1000): 4.53 -> 4.41 | Improved |
| Add affinity loss to binary cross-entropy mask loss (ablation) | CAPNet · ShapeNet | Chamfer distance | Improved |
| Use high affinity-loss weight throughout training (ablation) | CAPNet · ShapeNet chair | reconstruction of thin structures | Hurt |
| Use category-specific kernel variance instead of one multi-category setting (ablation) | CAPNet (multi-cat) · ShapeNet airplane | body-shape fidelity | Improved |
Closest papers by meaning
- Unsupervised Learning of Fine Structure Generation for 3D Point Clouds by 2D Projections Matching (ICCV 2021 · similarity 0.88)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.87)
- Multi-View Representation is What You Need for Point-Cloud Pre-Training (ICLR 2024 · similarity 0.87)
- APML: Adaptive Probabilistic Matching Loss for Robust 3D Point Cloud Reconstruction (NeurIPS 2025 · similarity 0.87)
- Hyperbolic-Constraint Point Cloud Reconstruction from Single RGB-D Images (AAAI 2025 · similarity 0.86)
- 3DMambaIPF: A State Space Model for Iterative Point Cloud Filtering via Differentiable Rendering (AAAI 2025 · 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.
