Research map / Point clouds / Point cloud classification
SampleNet: Differentiable Point Cloud Sampling
Itai Lang, Asaf Manor, Shai Avidan
Check 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 |
|---|---|---|---|
| train a task-specific sampler with differentiable soft projection, then sample from the input at inference | PointNet classification · ModelNet40 test, sampling ratio 32 | classification accuracy: 80.1% | Improved |
| use SampleNet instead of FPS for point-cloud registration | PCRNet · ModelNet40 Car test, sampling ratio 32 | mean rotation error: $13.46\degree$ -> $5.94\degree$ | Improved |
| use SampleNet instead of S-NET for point-cloud reconstruction | PCAE · ShapeNet Core55 test, sampling ratio 32 | normalized reconstruction error: 1.57 -> 1.33 | Improved |
| use a decaying temperature profile rather than a constant temperature (ablation) | SampleNet-Progressive classification · ModelNet40 test, sampling ratio 32 | classification accuracy: 17.1 -> 74.6 | Improved |
| use a linear rectified or exponential temperature schedule instead of the learned schedule (ablation) | SampleNet-Progressive classification · ModelNet40 test | classification accuracy | No clear effect |
| add cross-entropy loss to force projection weights toward the nearest neighbor (ablation) | SampleNet classification · ModelNet40 test, sampling ratio 32 | classification accuracy: 80.1 -> 74.4 | Hurt |
| add cross-entropy loss to force projection weights toward the nearest neighbor (ablation) | SampleNet classification · ModelNet40 test, sampling ratio 2 | classification accuracy: 88.4 -> 88.2 | Hurt |
| add entropy loss to force projection weights toward a one-hot distribution (ablation) | SampleNet classification · ModelNet40 test | classification accuracy | Hurt |
| use a projection neighborhood size other than k = 7 (ablation) | SampleNet-Progressive classification · ModelNet40 test | classification accuracy | Hurt |
| use FPS instead of SampleNet for airplane reconstruction | PCAE · ShapeNet Core55 airplane, sample size 64 | normalized reconstruction error: 1.39 -> 1.31 | Improved |
| use SampleNet for registration rather than FPS as sampling becomes sparser | PCRNet · ModelNet40 Car test, sampling ratio 32 | sampling consistency: 11.78 -> 6.85 | Improved |
Later papers that compared with this method
4 later accepted papers report SampleNet as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- Preserving Topological and Geometric Embeddings for Point Cloud Recovery (AAAI 2026 · better on 1 of 1 shared result)
- LTA-PCS: Learnable Task-Agnostic Point Cloud Sampling (CVPR 2024 · better on 6 of 6 shared results)
- Learnable Skeleton-Aware 3D Point Cloud Sampling (CVPR 2023 · better on 4 of 4 shared results)
- Meta-Sampler: Almost-Universal yet Task-Oriented Sampling for Point Clouds (ECCV 2022 · better on 4 of 4 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- GCD-Sampling: A General Cross-scale Decoupled Sampling for Point Cloud (AAAI 2025 · similarity 0.90)
- LTA-PCS: Learnable Task-Agnostic Point Cloud Sampling (CVPR 2024 · similarity 0.89)
- SAMBLE: Shape-Specific Point Cloud Sampling for an Optimal Trade-Off Between Local Detail and Global Uniformity (CVPR 2025 · similarity 0.88)
- MICAS: Multi-grained In-Context Adaptive Sampling for 3D Point Cloud Processing (CVPR 2025 · similarity 0.87)
- Point Cloud Pre-training with Diffusion Models (CVPR 2024 · similarity 0.87)
- Arbitrary-Scale Point Cloud Upsampling by Voxel-Based Network with Latent Geometric-Consistent Learning (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.
