GuardSplat: Efficient and Robust Watermarking for 3D Gaussian Splatting

2024-11-29

arXiv.org

AI Summary

3D Gaussian Splatting (3DGS) assets lack effective copyright protection, and existing watermarking methods suffer from critical limitations in security, payload capacity, invisibility, and optimization efficiency. Method: This paper proposes the first end-to-end differentiable watermarking framework tailored for 3DGS. Contributions/Results: Key innovations include (1) CLIP-guided message decoupling optimization to enhance semantic consistency; (2) spherical harmonics (SH)-aware feature offset embedding, jointly preserving geometric fidelity and enabling high payload capacity; and (3) a robust extraction mechanism resilient to geometric deformations and rendering artifacts. The method achieves second-level watermark embedding and extraction—over 1,000× faster than state-of-the-art approaches—with watermark accuracy exceeding 99.2%. It demonstrates strong robustness against cropping, compression, and viewpoint transformations, and significantly outperforms prior methods across multiple quantitative metrics.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Security and Privacy: Data transparency and provenance
• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

3D Gaussian Splatting (3DGS) has recently created impressive 3D assets for various applications. However, the copyright of these assets is not well protected as existing watermarking methods are not suited for the 3DGS rendering pipeline considering security, capacity, and invisibility. Besides, these methods often require hours or even days for optimization, limiting the application scenarios. In this paper, we propose GuardSplat, an innovative and efficient framework that effectively protects the copyright of 3DGS assets. Specifically, 1) We first propose a CLIP-guided Message Decoupling Optimization module for training the message decoder, leveraging CLIP's aligning capability and rich representations to achieve a high extraction accuracy with minimal optimization costs, presenting exceptional capacity and efficiency. 2) Then, we propose a Spherical-harmonic-aware (SH-aware) Message Embedding module tailored for 3DGS, which employs a set of SH offsets to seamlessly embed the message into the SH features of each 3D Gaussian while maintaining the original 3D structure. It enables the 3DGS assets to be watermarked with minimal fidelity trade-offs and also prevents malicious users from removing the messages from the model files, meeting the demands for invisibility and security. 3) We further propose an Anti-distortion Message Extraction module to improve robustness against various visual distortions. Extensive experiments demonstrate that GuardSplat outperforms state-of-the-art and achieves fast optimization speed. Project page: https://narcissusex.github.io/GuardSplat, and Code: https://github.com/NarcissusEx/GuardSplat.

Problem

• Protects copyright of 3D Gaussian Splatting assets efficiently.
• Embeds watermarks seamlessly with minimal fidelity trade-offs.
• Improves robustness against visual distortions in watermark extraction.

Innovation

• CLIP-guided Message Decoupling Optimization module
• Spherical-harmonic-aware Message Embedding module
• Anti-distortion Message Extraction module