Research map / Point clouds / Point cloud classification
PC-Adapter: Topology-Aware Adapter for Efficient Domain Adaption on Point Clouds with Rectified Pseudo-label
Joonhyung Park, Hyunjin Seo, Eunho Yang
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 locality-aware graph-convolution adapter to shape-aware adapter (ablation) | PC-Adapter · PointDA-10 S->M | accuracy: 72.3 -> 75.7 | Improved |
| add beta distribution-guided pseudo-labeling (Beta PL) (ablation) | PC-Adapter · PointDA-10 S->M | accuracy: 75.7 -> 77.5 | Improved |
| add beta distribution-guided pseudo-labeling (Beta PL) (ablation) | PC-Adapter · PointDA-10 S*->M | accuracy: 68.2 -> 73.7 | Improved |
| use proposed relative positional encoding instead of conventional encoding (ablation) | PC-Adapter · PointDA-10 S->M | accuracy: 74.2 -> 77.5 | Improved |
| use proposed relative positional encoding instead of conventional encoding (ablation) | PC-Adapter · PointDA-10 S*->M | accuracy: 70.3 -> 73.7 | Improved |
| use beta-guided pseudo-labels instead of maximum confidence pseudo-labels (ablation) | PC-Adapter · GraspNetPC-10 Syn.->RS. | balanced accuracy: 67.93 -> 71.99 | Improved |
| use beta-guided pseudo-labels instead of maximum confidence pseudo-labels (ablation) | PC-Adapter · GraspNetPC-10 Kin.->RS. | balanced accuracy: 79.49 -> 83.52 | Improved |
| train on farthest-point-sampled data to remove local structure, no adaptation (ablation) | PointNet and DGCNN · PointDA-10 multiple shifts | accuracy | Improved |
| PC-Adapter instead of best baseline on PointDA-10 | DGCNN · PointDA-10 | average accuracy: +5.3% | Improved |
| PC-Adapter on part segmentation with locality-aware adapter only | DGCNN · PointSegDA | average mIoU: 60.0 -> 63.5 | Improved |
| PC-Adapter on GraspNetPC-10 Syn.->RS. when combined with ImplicitPCDA | ImplicitPCDA · GraspNetPC-10 Syn.->RS. | accuracy: 80.5 -> 72.5 | Hurt |
| PC-Adapter instead of self-supervised baselines for training time | PC-Adapter · ModelNet->ShapeNet | training speed: 3.43x faster | Improved |
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.
- PointDAN — PointDAN: A Multi-Scale 3D Domain Adaption Network for Point Cloud Representation (NeurIPS 2019) · better on 2 of 3 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
- PointDAN: A Multi-Scale 3D Domain Adaption Network for Point Cloud Representation (NeurIPS 2019 · similarity 0.91)
- Domain Adaptation on Point Clouds via Geometry-Aware Implicits (CVPR 2022 · similarity 0.91)
- Self-Supervised Global-Local Structure Modeling for Point Cloud Domain Adaptation With Reliable Voted Pseudo Labels (CVPR 2022 · similarity 0.91)
- DAPoinTr: Domain Adaptive Point Transformer for Point Cloud Completion (AAAI 2025 · similarity 0.88)
- SMART-PC: Skeletal Model Adaptation for Robust Test-Time Training in Point Clouds (ICML 2025 · similarity 0.87)
- Density-guided Translator Boosts Synthetic-to-Real Unsupervised Domain Adaptive Segmentation of 3D Point Clouds (CVPR 2024 · similarity 0.86)
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.
