Research map / Adversarial robustness and security / Attacks on text and graphs
Generating Multi-view Adversarial Examples for Visual Geometry Grounded Transformer
Qi Song, Ziyuan Luo, Haoliang Han, Renjie Wan
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 single-view COCO pre-training (ablation) | MVAP-G generator · Component ablation | Chamfer distance: 0.58 -> 1.40 | Improved |
| add DPT head instead of Conv+Linear decoder (ablation) | MVAP-G generator · Component ablation | Chamfer distance: 1.40 -> 1.68 | Improved |
| add cross-view adversarial alignment (CAA) (ablation) | MVAP-G generator · Component ablation | Chamfer distance: 1.68 -> 1.75 | Improved |
| add CAA, adversarial loss (ablation) | MVAP-G generator · Component ablation | L_adv: 0.120 -> 0.008 | Improved |
| add regularization loss L_reg (ablation) | MVAP-G generator · Component ablation | PSNR: 20.3 -> 29.8 | Improved |
| add regularization loss L_reg (ablation) | MVAP-G generator · Component ablation | LPIPS: 0.512 -> 0.107 | Improved |
| add regularization loss L_reg (ablation) | MVAP-G generator · Component ablation | Chamfer distance: 1.75 -> 1.78 | No clear effect |
| add CAA, view count scaling (ablation) | MVAP-G generator · Fig (a) CD vs views | Chamfer distance | Improved |
| use generated perturbation (MVAP-G) instead of UAP | VGGT attack · CO3D (v=25) | shifted Chamfer distance: 0.659 -> 2.047 | Improved |
| use MVAP-G single pass instead of AP iter=20 | VGGT attack · COCO (v=1) | shifted Chamfer distance: 1.535 -> 1.566 | Improved |
| use MVAP-G instead of AP iter=20 | VGGT attack · LLFF (v=25) | shifted Chamfer distance: 1.925 -> 1.910 | Hurt |
| double AP iterations from 10 to 20 (ablation) | AP · COCO (v=1) | shifted Chamfer distance: 1.361 -> 1.535 | Improved |
Closest papers by meaning
- NeRFail: Neural Radiance Fields-Based Multiview Adversarial Attack (AAAI 2024 · similarity 0.86)
- RegimeVGGT: Layer-Wise Spatially Preserving Redundancy Removal for Visual Geometry Grounded Transformer (NeurIPS 2026 · similarity 0.86)
- Attacking Transformers with Feature Diversity Adversarial Perturbation (AAAI 2024 · similarity 0.86)
- VGGT: Visual Geometry Grounded Transformer (CVPR 2025 · similarity 0.86)
- Exploring Visual Vulnerabilities via Multi-Loss Adversarial Search for Jailbreaking Vision-Language Models (CVPR 2025 · similarity 0.86)
- G-NeRF: Geometry-enhanced Novel View Synthesis from Single-View Images (CVPR 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.
