Research map / Point clouds / Point cloud classification
Sample-adaptive Augmentation for Point Cloud Recognition Against Real-world Corruptions
Jie Wang, Lihe Ding, Tingfa Xu, Shaocong Dong, Xinli Xu, Long Bai, Jianan 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 |
|---|---|---|---|
| add AdaptPoint augmentation | RPC · ModelNet-C | mCE (%): 86.3 -> 56.5 | Improved |
| add AdaptPoint augmentation | RPC · ScanObjectNN-C | mCE (%): 132.6 -> 96.5 | Improved |
| add AdaptPoint augmentation | Point-MAE · ShapeNet-C part segmentation | mCE (%): 92.7 -> 76.9 | Improved |
| add AdaptPoint augmentation | DGCNN · ScanObjectNN clean | OA (%): 85.8 -> 84.4 | Hurt |
| add feedback loss (ablation) | AdaptPoint (PointNeXt) · ScanObjectNN-C | mCE (%): 82.1 -> 78.3 | Improved |
| add adversarial discriminator loss (ablation) | AdaptPoint (PointNeXt) · ScanObjectNN-C | mCE (%): 83.2 -> 78.3 | Improved |
| add both deformation and mask controllers (ablation) | AdaptPoint (PointNeXt) · ScanObjectNN-C | mCE (%): 92.1 -> 78.3 | Improved |
| add mask controller to deformation only (ablation) | AdaptPoint (PointNeXt) · ScanObjectNN-C | mCE (%): 86.4 -> 78.3 | Improved |
| increase anchors from 4 to 16 (ablation) | AdaptPoint deformation controller · ScanObjectNN-C | mCE (%): 78.3 -> 81.4 | Hurt |
| change feedback loss weight lambda from 1 to 0.5 or 2 (ablation) | AdaptPoint (PointNeXt) · ScanObjectNN-C | mCE (%): 78.3 -> 78.7 (0.5), 79.0 (2) | Hurt |
| apply AdaptPoint on PointNet++ for attack defense | PointNet++ · Point cloud attack defense Drop-200 | OA (%): 79.09 (IF-Defense) -> 83.59 | Improved |
| apply AdaptPoint instead of IF-Defense | PointNet++ · Point cloud attack defense kNN | OA (%): 85.62 -> 77.67 | Hurt |
Closest papers by meaning
- Target-Guided Adversarial Point Cloud Transformer Towards Recognition Against Real-world Corruptions (NeurIPS 2024 · similarity 0.88)
- ConcreTizer: Model Inversion Attack via Occupancy Classification and Dispersion Control for 3D Point Cloud Restoration (ICLR 2025 · similarity 0.87)
- PointMAC: Meta-Learned Adaptation for Robust Test-Time Point Cloud Completion (NeurIPS 2025 · similarity 0.87)
- SMART-PC: Skeletal Model Adaptation for Robust Test-Time Training in Point Clouds (ICML 2025 · similarity 0.86)
- Rethinking Point Cloud Data Augmentation: Topologically Consistent Deformation (ICML 2025 · similarity 0.86)
- Improving Robustness of 3D Point Cloud Recognition from a Fourier Perspective (NeurIPS 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.
