Research map / Point clouds / Shape and surface completion
SimNP: Learning Self-Similarity Priors Between Neural Points
Christopher Wewer, Eddy Ilg, Bernt Schiele, Jan Eric Lenssen
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 |
|---|---|---|---|
| remove shared point features S (ablation) | SimNP · ShapeNet cars, 1 view | PSNR: 23.00 -> 22.37 | Hurt |
| compute attention A via keys and queries instead of direct parameterization (ablation) | SimNP · ShapeNet cars, 1 view | PSNR: 23.00 -> 22.76 | No clear effect |
| increase number of embeddings from 128 to 1024 (ablation) | SimNP · ShapeNet cars, 1 view | PSNR: 23.09 -> 22.81 | Hurt |
| apply Gaussian blur to rendered images (ablation) | SimNP · ShapeNet cars, 1 view | PSNR: PSNR 23.00 -> 23.31; LPIPS 0.081 -> 0.092 | Mixed |
| add second input view from opposite side (ablation) | SimNP · ShapeNet cars | PSNR: 23.00 -> 26.69 | Improved |
| add depth supervision for point cloud prediction (ablation) | SimNP · ShapeNet cars, 1 view | PSNR: 23.00 -> 23.28 | Improved |
| add ray encodings, mask input and augmentation to point cloud encoder (ablation) | point cloud predictor · ShapeNet cars | Chamfer distance (x10^-3): 1.2103 -> 1.1028 | Improved |
| use SimNP on chairs versus VisionNeRF single view | SimNP · ShapeNet chairs, 1 view | PSNR: 24.48 -> 23.46 | Hurt |
| use SimNP instead of PixelNeRF on symmetric views of cars | SimNP · ShapeNet cars, 1 view (Sym.) | LPIPS: 0.122 -> 0.084 | Improved |
Closest papers by meaning
- Sharf: Shape-conditioned Radiance Fields from a Single View (ICML 2021 · similarity 0.88)
- GRF: Learning a General Radiance Field for 3D Representation and Rendering (ICCV 2021 · similarity 0.88)
- SNeS: Learning Probably Symmetric Neural Surfaces from Incomplete Data (ECCV 2022 · similarity 0.88)
- Neural Processing of Tri-Plane Hybrid Neural Fields (ICLR 2024 · similarity 0.86)
- UniSDF: Unifying Neural Representations for High-Fidelity 3D Reconstruction of Complex Scenes with Reflections (NeurIPS 2024 · similarity 0.86)
- 3D Reconstruction with Generalizable Neural Fields using Scene Priors (ICLR 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.
