Research map / Point clouds / Point cloud classification
PnP-3D: A Plug-and-Play for 3D Point Clouds
Shi Qiu, Saeed Anwar, Nick M. Barnes
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 PnP-3D module | PointNet++ · ScanObjectNN | overall accuracy: 77.9 -> 82.2 | Improved |
| add PnP-3D module | PointNet++ · ModelNet40 | overall accuracy: 90.7 -> 93.2 | Improved |
| add PnP-3D module | PAConv · ModelNet40 | overall accuracy: 93.6 -> 93.8 | Improved |
| add PnP-3D module | four segmentation baselines · S3DIS Area 5 | mIoU: 2.7% to 5.2% mIoU | Improved |
| add PnP-3D module | three detection baselines · SUN RGB-D V1 | mAP: 1.4% to 1.7% mAP | Improved |
| use average-pooling with subtraction instead of max-pooling with product (ablation) | PnP-3D on RandLA-Net · S3DIS Area 5 | mIoU: 63.0 -> 65.5 | Improved |
| use average-pooling instead of max-pooling (subtraction regularization) (ablation) | PnP-3D on RandLA-Net · S3DIS Area 5 | mIoU: 63.3 -> 65.5 | Improved |
| use geometric mean for global response instead of sum (ablation) | PnP-3D on RandLA-Net · S3DIS Area 5 | mIoU: 64.7 -> 65.5 | Improved |
| use geometric mean instead of grand mean for global response (ablation) | PnP-3D on RandLA-Net · S3DIS Area 5 | mIoU: 63.7 -> 65.5 | Improved |
| use PnP-3D instead of other attention modules (A-SCN) (ablation) | VoteNet · SUN RGB-D V1 | mAP@0.25: 55.6 -> 59.1 | Improved |
| use PnP-3D instead of Point-Transformer (ablation) | VoteNet · SUN RGB-D V1 | mAP@0.25: 58.1 -> 59.1 | Improved |
| use PnP-3D instead of additional RGB color input | VoteNet · SUN RGB-D V1 | mAP: 56.3 -> 59.1 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- SCF-Net — SCF-Net: Learning Spatial Contextual Features for Large-Scale Point Cloud Segmentation (CVPR 2021) · better on 1 of 1 shared result
- ImVoteNet — ImVoteNet: Boosting 3D Object Detection in Point Clouds With Image Votes (CVPR 2020) · better on 1 of 1 shared result
- RandLA-Net — RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds (CVPR 2020) · better on 1 of 1 shared result
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
- Starting From Non-Parametric Networks for 3D Point Cloud Analysis (CVPR 2023 · similarity 0.89)
- Learning of 3D Graph Convolution Networks for Point Cloud Analysis (TPAMI 2021 · similarity 0.88)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · similarity 0.87)
- Multi-View Representation is What You Need for Point-Cloud Pre-Training (ICLR 2024 · similarity 0.87)
- Point2Real: Bridging the Gap between Point Cloud and Realistic Image for Open-World 3D Recognition (AAAI 2024 · similarity 0.87)
- X-3D: Explicit 3D Structure Modeling for Point Cloud Recognition (CVPR 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.
