Research map / Point clouds / Point cloud classification
FLAT: Flux-aware Imperceptible Adversarial Attacks on 3D Point Clouds
Keke Tang, Lujie Huang, Weilong Peng, Daizong Liu, Xiaofei Wang, Yang Ma, Ligang Liu, Zhihong Tian
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 |
|---|---|---|---|
| suppress flux of perturbation vectors in local regions (FLAT) | point cloud attacks · ModelNet40 PointNet untargeted | Chamfer distance (1e-4): 2.747 (ITA) -> 1.539 | Improved |
| add flux-based uniformity preservation to IFGM (ablation) | IFGM · ModelNet40 PointNet | Chamfer distance (1e-4): 4.039 -> 2.944 | Improved |
| add flux-based uniformity preservation to ITA (ablation) | ITA · ModelNet40 PointNet | Hausdorff distance (1e-2): 0.414 -> 0.579 (CD 2.747 -> 2.253) | Mixed |
| use flux adjustment instead of direct SDM constraint on initial directions (ablation) | FLAT · ModelNet40 PointNet | Chamfer distance (1e-4): 2.502 -> 1.539 | Improved |
| use simplified flux instead of formal continuous-field flux (ablation) | FLAT · ModelNet40 PointNet | time per example (s): 150.07 -> 2.02 | Improved |
| choose initial perturbation direction from gradient (IFGM) vs tangent or normal (ablation) | FLAT · ModelNet40 PointNet | perturbation size | No clear effect |
| increase number of local regions up to 20 (ablation) | FLAT · ModelNet40 PointNet | l2 and EMD | Improved |
| increase local region radius beyond 0.1 (ablation) | FLAT · ModelNet40 PointNet | l2 and EMD | Hurt |
| apply defenses (SOR, DUP-Net) against FLAT | FLAT · PointNet | attack success rate | Hurt |
| transfer FLAT adversarial examples to other classifiers | FLAT · PointNet to DGCNN/PointConv | attack success rate | Hurt |
Closest papers by meaning
- 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 0.91)
- Imperceptible Transfer Attack and Defense on 3D Point Cloud Classification (TPAMI 2022 · similarity 0.90)
- Manifold Constraints for Imperceptible Adversarial Attacks on Point Clouds (AAAI 2024 · similarity 0.90)
- Imperceptible 3D Point Cloud Attacks on Lattice-based Barycentric Coordinates (AAAI 2025 · similarity 0.88)
- Explicitly Perceiving and Preserving the Local Geometric Structures for 3D Point Cloud Attack (AAAI 2024 · 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.
