GS-Checker: Tampering Localization for 3D Gaussian Splatting

2025-11-25

AI Summary

3D Gaussian Splatting (3DGS) models are vulnerable to malicious editing, yet no effective, annotation-free methods exist for localizing tampered regions in 3DGS reconstructions. Method: This paper proposes an unsupervised, 3D-annotation-free tampering localization framework. Its core innovations include: (i) explicitly embedding tampering attributes into learnable 3D Gaussian parameters—such as position, scale, and opacity—to construct a differentiable 3D tampering representation; (ii) designing a geometrically and photometrically consistent 3D contrastive mechanism to model local attribute similarity among Gaussian ellipsoids; and (iii) introducing a cyclic optimization strategy to enhance localization robustness. Results: Extensive experiments across diverse tampering scenarios demonstrate that our method significantly outperforms existing detection paradigms in both accuracy and generalization. It achieves fine-grained, pixel- and Gaussian-level tampering localization—constituting the first dedicated, end-to-end solution for securing 3DGS-rendered content.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Local Search
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Data transparency and provenance
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Recent advances in editing technologies for 3D Gaussian Splatting (3DGS) have made it simple to manipulate 3D scenes. However, these technologies raise concerns about potential malicious manipulation of 3D content. To avoid such malicious applications, localizing tampered regions becomes crucial. In this paper, we propose GS-Checker, a novel method for locating tampered areas in 3DGS models. Our approach integrates a 3D tampering attribute into the 3D Gaussian parameters to indicate whether the Gaussian has been tampered. Additionally, we design a 3D contrastive mechanism by comparing the similarity of key attributes between 3D Gaussians to seek tampering cues at 3D level. Furthermore, we introduce a cyclic optimization strategy to refine the 3D tampering attribute, enabling more accurate tampering localization. Notably, our approach does not require expensive 3D labels for supervision. Extensive experimental results demonstrate the effectiveness of our proposed method to locate the tampered 3DGS area.

Problem

• Detecting malicious tampering in 3D Gaussian Splatting models
• Localizing manipulated regions without expensive 3D supervision
• Identifying tampered 3D content using contrastive and optimization methods

Innovation

• Integrates tampering attribute into Gaussian parameters
• Designs 3D contrastive mechanism for tampering cues
• Uses cyclic optimization strategy without 3D supervision