Research map / 3D vision / 3D generation
GaussianDWM: 3D Gaussian Driving World Model for Unified Scene Understanding and Multi-Modal Generation
Tianchen Deng, Xuefeng Chen, Yi Chen, Qu Chen, Yuyao Xu, Lijin Yang, Le Xu, Yu Zhang, Bo Zhang, Wuxiong Huang, Hesheng Wang
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 3D Gaussian tokens to the fine-tuned LLM (ablation) | GaussianDWM · NuInteract | Avg score: 53.32 -> 55.21 (random sampling) | Improved |
| replace random sampling with top-k + uniform sampling of Gaussians (ablation) | GaussianDWM · NuInteract | Avg score: 55.21 -> 58.93 | Improved |
| add similarity-based language-guided sampling (ablation) | GaussianDWM · NuInteract | Avg score: 58.93 -> 59.23 | Mixed |
| fine-tune the model on NuInteract (ablation) | GaussianDWM · NuInteract | Avg score: 15.39 -> 53.32 | Improved |
| add high-level LLM language condition to low-level image condition (ablation) | GaussianDWM generator · nuScenes spatial generation | FID +-1m / +-2m: 10.12 -> 8.36 (+-1m); 45.14 -> 44.5 (+-2m) | Improved |
| use only high-level language condition without image condition (ablation) | GaussianDWM generator · nuScenes spatial generation | FID: failure | Hurt |
| GaussianDWM vs DriveMonkey on planning and region description | GaussianDWM · NuInteract | Planning Acc / CIDEr: 82.64 -> 80.95 Acc; 79.79 -> 78.72 CIDEr | Hurt |
| GaussianDWM vs DiST-S at large viewpoint shift | GaussianDWM · nuScenes spatial generation (+-4m) | FID / FVD: 17.57 -> 18.81 FID; 105.29 -> 116.40 FVD | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- OmniRe — OmniRe: Omni Urban Scene Reconstruction (ICLR 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- GaussianWorld: Gaussian World Model for Streaming 3D Occupancy Prediction (CVPR 2025 · similarity 0.88)
- Language Embedded 3D Gaussians for Open-Vocabulary Scene Understanding (CVPR 2024 · similarity 0.88)
- 3D Vision-Language Gaussian Splatting (ICLR 2025 · similarity 0.88)
- UniGS: Unified Language-Image-3D Pretraining with Gaussian Splatting (ICLR 2025 · similarity 0.88)
- MMScan: A Multi-Modal 3D Scene Dataset with Hierarchical Grounded Language Annotations (NeurIPS 2024 · similarity 0.88)
- Director3D: Real-world Camera Trajectory and 3D Scene Generation from Text (NeurIPS 2024 · similarity 0.88)
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.
