mlsanityData to 14 Oct 2026

StealthAttack: Robust 3D Gaussian Splatting Poisoning via Density-Guided Illusions

arXiv 2510.02314 ↗First posted 2 Oct 2025cs.CVcited by 6

Not on the map. Why?

Abstract (arXiv)

3D scene representation methods like Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) have significantly advanced novel view synthesis. As these methods become prevalent, addressing their vulnerabilities becomes critical. We analyze 3DGS robustness against image-level poisoning attacks and propose a novel density-guided poisoning method. Our method strategically injects Gaussian points into low-density regions identified via Kernel Density Estimation (KDE), embedding viewpoint-dependent illusory objects clearly visible from poisoned views while minimally affecting innocent views. Additionally, we introduce an adaptive noise strategy to disrupt multi-view consistency, further enhancing attack effectiveness. We propose a KDE-based evaluation protocol to assess attack difficulty systematically, enabling objective benchmarking for future research. Extensive experiments demonstrate our method's superior performance compared to state-of-the-art techniques. Project page: https://hentci.github.io/stealthattack/

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 88 of its 112 references that are in the map. First posted on arXiv 2 Oct 2025.

Builds on

88 of 112 references.

References · 88

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  4. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  5. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,226
  6. 2013Intriguing properties of neural networkscited by 5,155
  7. 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,723
  8. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
  9. 2021Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fieldscited by 2,032
  10. 2021Mip-NeRF: A Multiscale Representation for Anti-Aliasing Neural Radiance Fieldscited by 1,964
  11. 2016Adversarial examples in the physical worldcited by 1,769
  12. 2015DeepFool: a simple and accurate method to fool deep neural networkscited by 1,318
38 more
  1. 2020NeRF in the Wild: Neural Radiance Fields for Unconstrained Photo Collectionscited by 1,232
  2. 2021Efficient Geometry-aware 3D Generative Adversarial Networkscited by 1,205
  3. 2017Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoningcited by 1,055
  4. 2016Practical Black-Box Attacks against Machine Learningcited by 903
  5. 2023Dynamic 3D Gaussians: Tracking by Persistent Dynamic View Synthesiscited by 737
  6. 2019Square Attack: a query-efficient black-box adversarial attack via random searchcited by 685
  7. 2023Mip-Splatting: Alias-free 3D Gaussian Splattingcited by 667
  8. 2017ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Modelscited by 666
  9. 2023Nerfstudio: A Modular Framework for Neural Radiance Field Developmentcited by 600
  10. 2016Delving into Transferable Adversarial Examples and Black-box Attackscited by 536
  11. 2018Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networkscited by 519
  12. 2018Spectral Signatures in Backdoor Attackscited by 484
  13. 2018Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Surveycited by 460
  14. 2019Evading Defenses to Transferable Adversarial Examples by Translation-Invariant Attackscited by 442
  15. 2020Weight Poisoning Attacks on Pre-trained Modelscited by 341
  16. 2019HopSkipJumpAttack: A Query-Efficient Decision-Based Attackcited by 333
  17. 2017Certified Defenses for Data Poisoning Attackscited by 303
  18. 2023Text-to-3D using Gaussian Splattingcited by 299
  19. 2019Label-Consistent Backdoor Attackscited by 296
  20. 2020State of the Art on Neural Renderingcited by 278
  21. 2017Adversarial Examples for Semantic Segmentation and Object Detectioncited by 275
  22. 2018Adversarial Attacks and Defences: A Surveycited by 259
  23. 2019Simple Black-box Adversarial Attackscited by 251
  24. 2018Manipulating Machine Learning: Poisoning Attacks and Countermeasures for Regression Learningcited by 225
  25. 2024Mini-Splatting: Representing Scenes with a Constrained Number of Gaussianscited by 213
  26. 2024GaussianPro: 3D Gaussian Splatting with Progressive Propagationcited by 200
  27. 2017Towards Poisoning of Deep Learning Algorithms with Back-gradient Optimizationcited by 197
  28. 2023Robust Dynamic Radiance Fieldscited by 194
  29. 2021Unlearnable Examples: Making Personal Data Unexploitablecited by 193
  30. 2015DeepStereo: Learning to Predict New Views from the World's Imagerycited by 185
  31. 2016View Synthesis by Appearance Flowcited by 172
  32. 2023Adversarial Example Does Good: Preventing Painting Imitation from Diffusion Models via Adversarial Examplescited by 167
  33. 2023Raising the Cost of Malicious AI-Powered Image Editingcited by 163
  34. 2023HUGS: Human Gaussian Splatscited by 154
  35. 2018Generating 3D Adversarial Point Cloudscited by 144
  36. 2018Stronger Data Poisoning Attacks Break Data Sanitization Defensescited by 135
  37. 2017On the Robustness of Semantic Segmentation Models to Adversarial Attackscited by 126
  38. 2018Sever: A Robust Meta-Algorithm for Stochastic Optimizationcited by 125

50 most cited of 88 All 88 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 6.

6 papers.

  1. 2026Stroke of Surprise: Progressive Semantic Illusions in Vector Sketchingcited by 3
  2. 2026Spectral Defense Against Resource-Targeting Attack in 3D Gaussian Splattingcited by 1
  3. 2026PoInit-of-View: Poisoning Initialization of Views Transfers Across Multiple 3D Reconstruction Systemscited by 1
  4. 2026Characterizing Detectability in 3DGS Poisoning: A Stage-wise Benchmarkcited by 0
  5. 2026JanusMesh: Fast and Zero-Shot 3D Visual Illusion Generation via Cross-Space Denoisingcited by 0
  6. 2026A Survey on Adversarial Attacks and Defenses for Diffusion Models Across Multiple Modalitiescited by 0