Research map / Point clouds / Point cloud classification
Generating 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 |
|---|---|---|---|
| Shift existing points using optimized perturbations | PointNet · ModelNet40 targeted classification | success rate: 100% | Helped |
| Add independent points initialized at critical points and shifted by optimization | PointNet · ModelNet40 targeted classification | success rate: 100% | Helped |
| Use Hausdorff rather than Chamfer constraint for independent point generation | PointNet · ModelNet40 targeted classification | #points added: 88 vs 51 | Mixed |
| Add a second optimized adversarial cluster instead of one (ablation) | PointNet adversarial cluster attack · ModelNet40 targeted classification | success rate: 78.8% -> 98.2% | Helped |
| Add a third optimized adversarial cluster instead of two (ablation) | PointNet adversarial cluster attack · ModelNet40 targeted classification | success rate: 98.2% -> 99.3% | Helped |
| Add a second optimized object-shaped point cluster instead of one (ablation) | PointNet adversarial object attack · ModelNet40 targeted classification | success rate: 54.6% -> 93.8% | Helped |
| Add a third optimized object-shaped point cluster instead of two (ablation) | PointNet adversarial object attack · ModelNet40 targeted classification | success rate: 93.8% -> 97.3% | Helped |
| Transfer crafted attacks to PointNet++ rather than the attacked PointNet | PointNet-crafted attacks · Untargeted transfer, point shifting | success rate: 3.9% | Hurt |
| Transfer crafted attacks to DGCNN rather than the attacked PointNet | PointNet-crafted attacks · Untargeted transfer, 3 clusters | success rate: 16.9% | Hurt |
Closest papers by meaning
- Robust Adversarial Objects against Deep Learning Models (AAAI 2020 · similarity 0.94)
- Generating Transferable 3D Adversarial Point Cloud via Random Perturbation Factorization (AAAI 2023 · similarity 0.92)
- Hide in Thicket: Generating Imperceptible and Rational Adversarial Perturbations on 3D Point Clouds (CVPR 2024 · similarity 0.91)
- Curvature-Invariant Adversarial Attacks for 3D Point Clouds (AAAI 2024 · similarity 0.88)
- Explicitly Perceiving and Preserving the Local Geometric Structures for 3D Point Cloud Attack (AAAI 2024 · similarity 0.88)
- Seeing is Not Believing: Adversarial Natural Object Optimization for Hard-Label 3D Scene Attacks (CVPR 2025 · similarity 0.88)
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.