Research map / 3D vision / Gaussian splatting
GaussTrace: Provenance Analysis of 3D Gaussian Splatting Models with Evidence-based LLM Reasoning
Haoliang Han, Ziyuan Luo, Renjie Wan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the chain-of-thought reasoning protocol (ablation) | GaussTrace · 3DGS provenance dataset | Edge Overlap: 0.811 -> 0.890 | Improved |
| add the chain-of-thought reasoning protocol (ablation) | GaussTrace · 3DGS provenance dataset | Vertex Edge Overlap: 0.913 -> 0.948 | Improved |
| add the chain-of-thought reasoning protocol (ablation) | GaussTrace · 3DGS provenance dataset | Vertex Overlap: 1.000 -> 1.000 | No clear effect |
| replace rule-based search with LLM reasoning | Rule-based search · 3DGS provenance dataset | Edge Overlap: 0.319 -> 0.890 | Improved |
| replace ViT-B+Integrity with GaussTrace | ViT-B+Integrity · 3DGS provenance dataset | Edge Overlap: 0.470 -> 0.890 | Improved |
| add 10% distractor models | GaussTrace · 3DGS provenance dataset | Edge Overlap: 0.890 -> 0.811 | Hurt |
| add 20% distractor models | GaussTrace · 3DGS provenance dataset | Edge Overlap: 0.890 -> 0.795 | Hurt |
| add 20% distractor models | GaussTrace · 3DGS provenance dataset | Vertex Overlap: 1.000 -> 0.940 | Hurt |
| add each Gaussian attribute group (position, opacity, scale, rotation, color) to the signature (ablation) | GaussTrace · 3DGS provenance dataset | Edge Overlap | Improved |
| swap GPT-5 for Qwen3-Max | GaussTrace · counter scene | Edge Overlap: 1.000 -> 0.857 | Hurt |
| swap GPT-5 for Llama-4 | GaussTrace · counter scene | Edge Overlap: 1.000 -> 0.933 | Hurt |
| swap GPT-5 for Gemini-2.5-Pro | GaussTrace · counter scene | Edge Overlap: 1.000 -> 1.000 | No clear effect |
Closest papers by meaning
- AG2aussian: Anchor-Graph Structured Gaussian Splatting for Instance-Level 3D Scene Understanding and Editing (ICCV 2025 · similarity 0.84)
- Witness Overlap: Directional Provenance Inside Open-Weight Model Families (NeurIPS 2026 · similarity 0.84)
- GaussianMarker: Uncertainty-Aware Copyright Protection of 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.83)
- TRACER: Verifiable Generative Provenance for Multimodal Tool-Using Agents (NeurIPS 2026 · similarity 0.83)
- Model Provenance Testing for Large Language Models (NeurIPS 2025 · similarity 0.82)
- FlexGS: Train Once, Deploy Everywhere with Many-in-One Flexible 3D Gaussian Splatting (CVPR 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
