Research map / Point clouds / Point cloud classification
Shape-Invariant 3D Adversarial Point Clouds
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| constrain point perturbations to tangent planes and rank points by sensitivity | white-box point-cloud attack · ModelNet40, PointNet / DGCNN / CurveNet | Chamfer distance: 2.15; 6.33; 6.25 | Helped |
| use tangent-plane sensitivity ranking to select attack points | query-based attack · ModelNet40, PointNet++ | average queries: 987.9 -> 417.0 | Helped |
| perturb points in descending rather than ascending or random sensitivity order | sensitivity-map attack on PAConv · ModelNet40 | recognition accuracy degradation | Helped |
| use the proposed tangent-constrained sensitivity-map attack instead of SimBA+ | black-box attack on PointNet++ · ModelNet40 | average queries: 603.4 -> 417.0 | Helped |
| use the proposed tangent-constrained sensitivity-map attack instead of SimBA+ | black-box attack on PAConv · ModelNet40 | average queries: 67.4 -> 53.9 | Helped |
| use the proposed tangent-constrained sensitivity-map attack instead of SimBA+ | black-box attack on CurveNet · ModelNet40 | average queries: 98.9 -> 81.5 | Helped |
| increase black-box attack step size (ablation) | shape-invariant query-based attack · ModelNet40, PAConv | average queries | Mixed |
| use PointNet instead of DGCNN as the white-box surrogate (ablation) | black-box attack · ModelNet40, SimpleView / PAConv / CurveNet | attack success rate: 100.0 -> 98.6; 100.0 -> 99.7; 99.9 -> 98.9 | Mixed |
| increase point dropout from 30% to 50% (ablation) | black-box attack on PAConv · ModelNet40 | ASR (%): 75.4 -> 84.4 | Helped |
| increase point dropout from 30% to 50% (ablation) | black-box attack on SimpleView · ModelNet40 | ASR (%): 51.1 -> 63.7 | Helped |
Closest papers by meaning
- Curvature-Invariant Adversarial Attacks for 3D Point Clouds (AAAI 2024 · similarity 0.93)
- Hide in Thicket: Generating Imperceptible and Rational Adversarial Perturbations on 3D Point Clouds (CVPR 2024 · similarity 0.93)
- Explicitly Perceiving and Preserving the Local Geometric Structures for 3D Point Cloud Attack (AAAI 2024 · similarity 0.90)
- Manifold Constraints for Imperceptible Adversarial Attacks on Point Clouds (AAAI 2024 · similarity 0.90)
- Improving Robustness of 3D Point Cloud Recognition from a Fourier Perspective (NeurIPS 2024 · similarity 0.89)
- PointCVaR: Risk-Optimized Outlier Removal for Robust 3D Point Cloud Classification (AAAI 2024 · similarity 0.89)
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.