Research map / Point clouds / Point cloud classification
Frugal 3D Point Cloud Model Training via Progressive Near Point Filtering and Fused Aggregation
Donghyun Lee, Yejin Lee, Jae W. Lee, Hongil Yoon
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 |
|---|---|---|---|
| replace FPS with Lightweight FPS using progressive near point filtering | PointNet-based models · S3DIS and ScanNet | mIoU (%): max mIoU loss of 0.06%, potential gain of 0.2% | No clear effect |
| add L-FPS only (ablation) | PointNet-based models · S3DIS and ScanNet (geomean) | end-to-end training speedup: 1.72x | Improved |
| add fused aggregation only (ablation) | PointNet-based models · S3DIS and ScanNet (geomean) | end-to-end training speedup: 1.22x | Improved |
| combine L-FPS and fused aggregation (ablation) | PointNet-based models · S3DIS and ScanNet (geomean) | end-to-end training speedup: 2.25x | Improved |
| apply both techniques, GPU energy | PointNet-based models · S3DIS and ScanNet (geomean) | energy efficiency: 1.94x | Improved |
| apply L-FPS on ModelNet40 with 1024 points per scene | PointNet++ · ModelNet40 | speedup: 1.16x slowdown | Hurt |
| apply fused aggregation on ModelNet40 | PointNet++ · ModelNet40 | speedup: 1.22x | Improved |
| use smaller training batch size (ablation) | L-FPS plus fused aggregation · S3DIS and ScanNet | speedup | Improved |
| scale model size from PointNet++ to PointMetaBase-XL (ablation) | L-FPS versus fused aggregation · S3DIS | speedup contribution | Mixed |
Closest papers by meaning
- 3DMambaIPF: A State Space Model for Iterative Point Cloud Filtering via Differentiable Rendering (AAAI 2025 · similarity 0.86)
- ComPC: Completing a 3D Point Cloud with 2D Diffusion Priors (ICLR 2025 · similarity 0.85)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.85)
- PLATYPUS: Progressive Local Surface Estimator for Arbitrary-Scale Point Cloud Upsampling (AAAI 2025 · similarity 0.85)
- GCD-Sampling: A General Cross-scale Decoupled Sampling for Point Cloud (AAAI 2025 · similarity 0.85)
- StraightPCF: Straight Point Cloud Filtering (CVPR 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.
