Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2504.20829 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912353606434816 |
|---|---|
| author | Hong, Jiaxin Chen, Sixu Sun, Shuoyang Yu, Hongyao Fang, Hao Tan, Yuqi Chen, Bin Qi, Shuhan Li, Jiawei |
| author_facet | Hong, Jiaxin Chen, Sixu Sun, Shuoyang Yu, Hongyao Fang, Hao Tan, Yuqi Chen, Bin Qi, Shuhan Li, Jiawei |
| contents | As 3D Gaussian Splatting (3DGS) emerges as a breakthrough in scene representation and novel view synthesis, its rapid adoption in safety-critical domains (e.g., autonomous systems, AR/VR) urgently demands scrutiny of potential security vulnerabilities. This paper presents the first systematic study of backdoor threats in 3DGS pipelines. We identify that adversaries may implant backdoor views to induce malicious scene confusion during inference, potentially leading to environmental misperception in autonomous navigation or spatial distortion in immersive environments. To uncover this risk, we propose GuassTrap, a novel poisoning attack method targeting 3DGS models. GuassTrap injects malicious views at specific attack viewpoints while preserving high-quality rendering in non-target views, ensuring minimal detectability and maximizing potential harm. Specifically, the proposed method consists of a three-stage pipeline (attack, stabilization, and normal training) to implant stealthy, viewpoint-consistent poisoned renderings in 3DGS, jointly optimizing attack efficacy and perceptual realism to expose security risks in 3D rendering. Extensive experiments on both synthetic and real-world datasets demonstrate that GuassTrap can effectively embed imperceptible yet harmful backdoor views while maintaining high-quality rendering in normal views, validating its robustness, adaptability, and practical applicability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_20829 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | GaussTrap: Stealthy Poisoning Attacks on 3D Gaussian Splatting for Targeted Scene Confusion Hong, Jiaxin Chen, Sixu Sun, Shuoyang Yu, Hongyao Fang, Hao Tan, Yuqi Chen, Bin Qi, Shuhan Li, Jiawei Computer Vision and Pattern Recognition Artificial Intelligence As 3D Gaussian Splatting (3DGS) emerges as a breakthrough in scene representation and novel view synthesis, its rapid adoption in safety-critical domains (e.g., autonomous systems, AR/VR) urgently demands scrutiny of potential security vulnerabilities. This paper presents the first systematic study of backdoor threats in 3DGS pipelines. We identify that adversaries may implant backdoor views to induce malicious scene confusion during inference, potentially leading to environmental misperception in autonomous navigation or spatial distortion in immersive environments. To uncover this risk, we propose GuassTrap, a novel poisoning attack method targeting 3DGS models. GuassTrap injects malicious views at specific attack viewpoints while preserving high-quality rendering in non-target views, ensuring minimal detectability and maximizing potential harm. Specifically, the proposed method consists of a three-stage pipeline (attack, stabilization, and normal training) to implant stealthy, viewpoint-consistent poisoned renderings in 3DGS, jointly optimizing attack efficacy and perceptual realism to expose security risks in 3D rendering. Extensive experiments on both synthetic and real-world datasets demonstrate that GuassTrap can effectively embed imperceptible yet harmful backdoor views while maintaining high-quality rendering in normal views, validating its robustness, adaptability, and practical applicability. |
| title | GaussTrap: Stealthy Poisoning Attacks on 3D Gaussian Splatting for Targeted Scene Confusion |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2504.20829 |