mlsanityData to 7 Oct 2026

Generating Multi-view Adversarial Examples for Visual Geometry Grounded Transformer

arXiv 2608.20748 ↗First posted 21 Aug 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

The Visual Geometry Grounded Transformer (VGGT) enables unified feed-forward 3D reconstruction from multi-view images. However, deploying such a high-performance model may expose critical security vulnerabilities. Traditional adversarial perturbations require costly per-scene optimization, while Universal Adversarial Perturbations (UAPs) rely on a single static pattern and fail to effectively attack VGGT. To address these limitations, we propose MVAP-G, a multi-view adversarial perturbation generator that produces imperceptible consistent perturbations across multiple views in a single feed-forward pass. To ensure perturbation consistency across diverse scenes, we design a cross-view adversarial alignment mechanism to process multi-view images. Experiments demonstrate that MVAP-G significantly degrades VGGT performance without iterative optimization during inference. This work pioneers multi-view adversarial attacks on 3D foundation models, uncovering severe vulnerabilities and underscoring the urgent need for robust 3D vision systems. The code is available at https://github.com/qsong2001/mvap-g.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 51 of its 70 references that are in the map. First posted on arXiv 21 Aug 2026.

Builds on

51 of 70 references.

References · 51

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  4. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,029
  5. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,764
  6. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  7. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,893
  8. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  9. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,141
  10. 2017ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenescited by 3,088
  11. 2021Vision Transformers for Dense Predictioncited by 1,789
  12. 2025VGGT: Visual Geometry Grounded Transformercited by 1,475
38 more
  1. 2024Depth Anything V2cited by 1,385
  2. 2024Grounding Image Matching in 3D with MASt3Rcited by 960
  3. 2015A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimationcited by 878
  4. 2016Universal adversarial perturbationscited by 774
  5. 2021MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereocited by 735
  6. 2021IBRNet: Learning Multi-View Image-Based Renderingcited by 698
  7. 2021Common Objects in 3D: Large-Scale Learning and Evaluation of Real-life 3D Category Reconstructioncited by 646
  8. 2019BlendedMVS: A Large-scale Dataset for Generalized Multi-view Stereo Networkscited by 487
  9. 2019Nesterov Accelerated Gradient and Scale Invariance for Adversarial Attackscited by 352
  10. 2016Virtual Worlds as Proxy for Multi-Object Tracking Analysiscited by 340
  11. 2017Adversarial Examples for Semantic Segmentation and Object Detectioncited by 275
  12. 2013Maxout Networkscited by 186
  13. 2018ShapeShifter: Robust Physical Adversarial Attack on Faster R-CNN Object Detectorcited by 159
  14. 2017Generative Adversarial Perturbationscited by 153
  15. 2025VGGT-SLAM: Dense RGB SLAM Optimized on the SL(4) Manifoldcited by 130
  16. 2025VGGT-Long: Chunk it, Loop it, Align it -- Pushing VGGT's Limits on Kilometer-scale Long RGB Sequencescited by 107
  17. 2025FastVGGT: Training-Free Acceleration of Visual Geometry Transformercited by 93
  18. 2025SpatialTrackerV2: 3D Point Tracking Made Easycited by 84
  19. 2017Adversarial Transformation Networks: Learning to Generate Adversarial Examplescited by 60
  20. 2017NO Need to Worry about Adversarial Examples in Object Detection in Autonomous Vehiclescited by 48
  21. 2017Geometric robustness of deep networks: analysis and improvementcited by 45
  22. 2023CopyRNeRF: Protecting the CopyRight of Neural Radiance Fieldscited by 33
  23. 2024DarkSAM: Fooling Segment Anything Model to Segment Nothingcited by 32
  24. 2025X-Transfer Attacks: Towards Super Transferable Adversarial Attacks on CLIPcited by 21
  25. 2024Protecting NeRFs' Copyright via Plug-And-Play Watermarking Base Modelcited by 18
  26. 2024Poison-splat: Computation Cost Attack on 3D Gaussian Splattingcited by 17
  27. 2024Disrupting Diffusion: Token-Level Attention Erasure Attack against Diffusion-based Customizationcited by 15
  28. 2024Geometry Cloak: Preventing TGS-based 3D Reconstruction from Copyrighted Imagescited by 15
  29. 2023NeRFool: Uncovering the Vulnerability of Generalizable Neural Radiance Fields against Adversarial Perturbationscited by 10
  30. 2023Targeted Adversarial Attacks on Generalizable Neural Radiance Fieldscited by 10
  31. 2024Texture Re-scalable Universal Adversarial Perturbationcited by 9
  32. 2020Learning to Generate Image Source-Agnostic Universal Adversarial Perturbationscited by 7
  33. 2024Towards Million-Scale Adversarial Robustness Evaluation With Stronger Individual Attackscited by 3
  34. 2025Improving Generalization of Universal Adversarial Perturbation via Dynamic Maximin Optimizationcited by 3
  35. 2025GS-Checker: Tampering Localization for 3D Gaussian Splattingcited by 3
  36. 2026Semantic-aware Adversarial Fine-tuning for CLIPcited by 2
  37. 2025Unified Steganography via Implicit Neural Representationcited by 1
  38. 2025Creating Blank Canvas Against AI-enabled Image Forgerycited by 1

50 most cited of 51 All 51 (JSON)

Built on by

Not cited yet.