Research map / Point clouds / Point cloud classification
PointVector: A Vector Representation in Point Cloud Analysis
Xin Deng, WenYu Zhang, Qing Ding, XinMing Zhang
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 PointNeXt's InvResMLP with VPSA vector-oriented aggregation | PointNeXt-XL · S3DIS 6-fold cross-validation | mIOU: 74.9 -> 78.4 | Improved |
| Replace PointNeXt's InvResMLP with VPSA vector-oriented aggregation | PointNeXt-XL · S3DIS Area 5 | mIOU: 70.8 -> 72.3 | Improved |
| Replace PointNeXt-S aggregation with vector-oriented aggregation | PointNeXt-S · ScanObjectNN PB_T50_RS | OA: 87.7±0.4 -> 87.8±0.4 | No clear effect |
| Replace PointNeXt-S aggregation with vector-oriented aggregation | PointNeXt-S · ScanObjectNN PB_T50_RS | mAcc: 85.8±0.6 -> 86.2±0.5 | Improved |
| Use 3D rather than scalar representations (ablation) | PointVector-L · S3DIS Area 5 | mIOU: 69.8 -> 71.2 | Improved |
| Use 3D rather than 2D vector representations (ablation) | PointVector-L · S3DIS Area 5 | mIOU: 70.9 -> 71.2 | Improved |
| Generate vector representations by rotation-based expansion rather than MLP (ablation) | PointVector-L · S3DIS Area 5 | mIOU: 70.8 -> 71.2 | Improved |
| Use sum rather than max reduction with GroupConv (ablation) | PointVector-L · S3DIS Area 5 | mIOU: 70.6 -> 71.2 | Improved |
| Use fixed convolution instead of sum+GroupConv (ablation) | PointVector-L · S3DIS Area 5 | mIOU: 71.2 -> 69.4 | Hurt |
| Replace vector encoder and vector aggregation with MLP+max+MLPs (ablation) | PointVector-XL · S3DIS Area 5 | mIOU: 71.4 -> 72.3 | Hurt |
| Add relative features and positions instead of concatenating them (ablation) | PointVector-L · S3DIS Area 5 | mIOU: 64.9 -> 71.2 | Improved |
| Use relative rather than absolute features (ablation) | PointVector-L · S3DIS Area 5 | mIOU: 70.5 -> 71.2 | Improved |
Later papers that compared with this method
1 later accepted paper reports PointVector-XL as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- X-3D: Explicit 3D Structure Modeling for Point Cloud Recognition (CVPR 2024 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PointNeXt-XL — PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategies (NeurIPS 2022) · better on 2 of 2 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
- Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework (ICLR 2022 · similarity 0.88)
- A Linear Time and Space Local Point Cloud Geometry Encoder via Vectorized Kernel Mixture (VecKM) (ICML 2024 · similarity 0.88)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · similarity 0.87)
- Point Transformer V3: Simpler Faster Stronger (CVPR 2024 · similarity 0.87)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.87)
- Improved MLP Point Cloud Processing with High-Dimensional Positional Encoding (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.
