Research map / Point clouds / Point cloud classification
Deformation-based In-Context Learning for Point Cloud Understanding
Chengxing Lin, Jinhong Deng, Yinjie Lei, Wen Li
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 |
|---|---|---|---|
| fully mask the query target during training to match inference (ablation) | PIC-Cat · ShapeNet In-Context ablation | Registration CD: 14.1 -> 3.5 | Improved |
| fully mask the query target during training (ablation) | PIC-Cat · ShapeNet In-Context ablation | Reconstruction CD: 4.3 -> 3.2 | Improved |
| add Deformation Transfer Network (deformation instead of masked reconstruction) (ablation) | PIC-Cat + consistency · ShapeNet In-Context ablation | Registration CD: 3.5 -> 2.9 | Improved |
| add Deformation Extraction Network (ablation) | PIC-Cat + DTN · ShapeNet In-Context ablation | Registration CD: 2.9 -> 2.0 | Improved |
| add Deformation Extraction Network (ablation) | PIC-Cat + DTN · ShapeNet In-Context ablation | Denoising CD: 3.8 -> 3.5 | Improved |
| replace masked point modeling with deformation | PIC-S-Sep · ShapeNet In-Context | Registration CD: 6.7 -> 2.0 | Improved |
| use DeformPIC instead of multitask Point-MAE out of domain | Point-MAE · ModelNet40 In-Context | Registration CD: 38.2 -> 2.3 | Improved |
| use DeformPIC instead of specialized ACT on reconstruction L5 | ACT (task-specific) · ShapeNet In-Context L5 | Reconstruction CD: 2.8 -> 3.9 | Hurt |
| use DeformPIC instead of PIC-S-Cat on part segmentation | PIC-S-Cat · ShapeNet In-Context | Part segmentation mIoU: 83.8 -> 83.9 | Improved |
| use DeformPIC instead of PIC-Cat on ScanObjectNN denoising | PIC-Cat · ScanObjectNN In-Context | Denoising CD: 9.2 -> 5.0 | Improved |
Closest papers by meaning
- Explore In-Context Learning for 3D Point Cloud Understanding (NeurIPS 2023 · similarity 0.90)
- Point Deformable Network with Enhanced Normal Embedding for Point Cloud Analysis (AAAI 2024 · similarity 0.88)
- Point-Cache: Test-time Dynamic and Hierarchical Cache for Robust and Generalizable Point Cloud Analysis (CVPR 2025 · similarity 0.86)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · similarity 0.86)
- Rethinking Point Cloud Data Augmentation: Topologically Consistent Deformation (ICML 2025 · similarity 0.86)
- MICAS: Multi-grained In-Context Adaptive Sampling for 3D Point Cloud Processing (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.
