Research map / Point clouds / Point cloud classification
TopAdapter: Topology-Aware Prompt Tuning for Efficient Point Cloud Understanding
Changshuo Wang, Shuting He, Xiang Fang, Weijun Li, Yixian Shen, Mingkun Xu, Zhongtian Sun, Prayag Tiwari
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 Topology Injection module (ToInjection) alone (ablation) | PointGPT-L · ScanObjectNN PB_T50_RS | accuracy (%): 93.40 -> 94.26 | Improved |
| add Topology Transfer module (ToTransfer) alone (ablation) | PointGPT-L · ScanObjectNN PB_T50_RS | accuracy (%): 93.40 -> 94.13 | Improved |
| combine ToInjection and ToTransfer (ablation) | TopAdapter on PointGPT-L · ScanObjectNN PB_T50_RS | accuracy (%): 94.26 -> 94.97 | Improved |
| add 0D, 1D and 2D simplex features over no simplex (ablation) | TopAdapter on PointGPT-L · ScanObjectNN PB_T50_RS | accuracy (%): 93.62 -> 94.97 | Improved |
| use two simplex dimensions instead of one (ablation) | TopAdapter on PointGPT-L · ScanObjectNN PB_T50_RS | accuracy (%): 94.21 -> 94.61 | Improved |
| set geometric control parameter alpha to 0.6 (ablation) | TopAdapter on PointGPT-L · ScanObjectNN PB_T50_RS | accuracy (%): 94.28 (alpha 0.1) -> 94.97 (alpha 0.6) -> 94.52 (alpha 0.9) | Mixed |
| set neighbor count K to 30 (ablation) | TopAdapter on PointGPT-L · ScanObjectNN PB_T50_RS | accuracy (%): 93.82 (K=10) -> 94.97 (K=30) | Mixed |
| apply TopAdapter instead of full fine-tuning | PointGPT-L · ScanObjectNN PB_T50_RS | accuracy (%): 93.40 -> 94.97 | Improved |
| apply TopAdapter on ReCon for ShapeNetPart segmentation | ReCon · ShapeNetPart | class mIoU (%): 84.52 -> 83.99 | Hurt |
| apply TopAdapter on ReCon for ScanObjectNN PB_T50_RS | ReCon · ScanObjectNN PB_T50_RS | accuracy (%): 89.73 -> 89.82 | No clear effect |
| use general PEFT methods (Adapter, LoRA) instead of point-cloud-specific ones | Point-MAE · ScanObjectNN PB_T50_RS | accuracy (%): 81.74 (LoRA) vs 85.58 (TopAdapter) | Hurt |
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.
- GAPrompt — GAPrompt: Geometry-Aware Point Cloud Prompt for 3D Vision Model (ICML 2025) · better on 3 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
- GAPrompt: Geometry-Aware Point Cloud Prompt for 3D Vision Model (ICML 2025 · similarity 0.89)
- On Geometry-Enhanced Parameter-Efficient Fine-Tuning for 3D Scene Segmentation (NeurIPS 2025 · similarity 0.88)
- Point-PEFT: Parameter-Efficient Fine-Tuning for 3D Pre-trained Models (AAAI 2024 · similarity 0.88)
- Point Transformer V3: Simpler Faster Stronger (CVPR 2024 · similarity 0.87)
- GTA: A Geometry-Aware Attention Mechanism for Multi-View Transformers (ICLR 2024 · similarity 0.86)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · 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.
