Research map / Point clouds / Shape and surface completion
Adaptive Surface Reconstruction With Multiscale Convolutional Kernels
Benjamin Ummenhofer, Vladlen Koltun
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 the normal loss L_Normal (ablation) | multiscale-kernel ConvNet · Tanks and Temples (7 scenes) | mean F-score: 43.96 -> 45.15 | Improved |
| add COLMAP point clouds from rendered images to training data (ablation) | multiscale-kernel ConvNet · Tanks and Temples (7 scenes) | mean F-score: 44.87 -> 45.15 | Improved |
| fine-tune Points2Surf on the authors' training data | Points2Surf · Tanks and Temples (7 scenes) | mean F-score: 18.61 -> 23.13 | Improved |
| replace PSR with the learned multiscale ConvNet | PSR · Tanks and Temples (7 scenes) | mean F-score: 40.09 -> 45.15 | Improved |
| replace GDMR with the learned multiscale ConvNet on Citywall | GDMR · Citywall (359M points) | runtime (min): 274 -> 117 | Improved |
| use the multiscale ConvNet instead of PSR on Ignatius | PSR · Tanks and Temples Ignatius | F-score: 75.58 vs 76.62 | No clear effect |
| use the multiscale ConvNet instead of PSR on Courthouse | PSR · Tanks and Temples Courthouse | F-score: 16.47 -> 18.41 | Improved |
| fine-tune Points2Surf on the authors' training data (Truck) | Points2Surf · Tanks and Temples Truck | F-score: 15.81 -> 33.71 | Improved |
| fine-tune Points2Surf on the authors' training data (Barn) | Points2Surf · Tanks and Temples Barn | F-score: 16.69 -> 18.33 | Mixed |
| fine-tune Points2Surf on the authors' training data (Ignatius) | Points2Surf · Tanks and Temples Ignatius | F-score: 50.87 -> 43.75 | Hurt |
Closest papers by meaning
- Neural Kernel Surface Reconstruction (CVPR 2023 · similarity 0.90)
- VKSR: Scalable Kernel Surface Reconstruction Using Vecchia's Approximation (ECCV 2026 · similarity 0.88)
- Matérn Kernels for Tunable Implicit Surface Reconstruction (ICLR 2025 · similarity 0.87)
- Arbitrary-Scale Point Cloud Upsampling by Voxel-Based Network with Latent Geometric-Consistent Learning (AAAI 2024 · similarity 0.86)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.86)
- GVKF: Gaussian Voxel Kernel Functions for Highly Efficient Surface Reconstruction in Open Scenes (NeurIPS 2024 · similarity 0.85)
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.
