Research map / 3D vision / Gaussian splatting
Street Gaussians: Modeling Dynamic Urban Scenes with Gaussian Splatting
Yunzhi Yan, Haotong Lin, Chenxu Zhou, Weijie Wang, Haiyang Sun, Kun Zhan, Xianpeng Lang, Xiaowei Zhou, Sida Peng
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 tracked pose optimization (ablation) | Street Gaussians · Waymo Open, moving objects | PSNR*: 28.24 -> 30.23 | Improved |
| add 4D spherical harmonics for object appearance (ablation) | Street Gaussians · Waymo Open, moving objects | PSNR*: 29.27 -> 30.23 | Improved |
| initialize with aggregated LiDAR points and add LiDAR depth loss (ablation) | Street Gaussians · Waymo Open | PSNR: 34.02 -> 34.61 | Improved |
| combine SfM points with LiDAR points for background init (ablation) | Street Gaussians · Waymo (2 large-scale scenes) | PSNR: 29.97 -> 32.63 | Improved |
| model sky with a separate cubemap (ablation) | Street Gaussians · Waymo (2 large-scale scenes) | PSNR: 31.12 -> 32.63 | Improved |
| optimize tracked poses instead of using ground-truth poses (ablation) | Street Gaussians · Waymo Open, moving objects | PSNR*: 29.84 -> 30.23 | Improved |
| pose optimization on implicit MARS representation versus explicit Gaussians (ablation) | MARS · Waymo sequence A | PSNR*: MARS 27.68 vs GT 29.74; Ours 31.35 vs GT 30.84 | Mixed |
| replace NeRF-based dynamic scene baselines with explicit Street Gaussians | Waymo rendering · Waymo Open | PSNR: 30.87 -> 34.61 | Improved |
| use Street Gaussians instead of static 3D GS for speed | Waymo rendering · Waymo Open | FPS: 205 -> 135 | Hurt |
| fuse semantic logits in 3D instead of per-frame Video K-Net predictions | Street Gaussians · KITTI segmentation | mIoU: 53.81 -> 58.81 | Improved |
Later papers that compared with this method
11 later accepted papers report Street Gaussians as a baseline in their results tables; 10 of them report a better number on most of the results they share with it.
- UFO: Unifying Feed-Forward and Optimization-based Methods for Large Driving Scene Modeling (CVPR 2026 · better on 1 of 1 shared result)
- LidarPainter: One-Step Away from Any Lidar View to Novel Guidance (AAAI 2026 · better on 1 of 1 shared result)
- ReinDriveGen: Reinforcement Post-Training for Out-of-Distribution Driving Scene Generation (ECCV 2026 · better on 1 of 1 shared result)
- ReconDreamer++: Harmonizing Generative and Reconstructive Models for Driving Scene Representation (ICCV 2025 · better on 8 of 8 shared results)
- ReconDreamer: Crafting World Models for Driving Scene Reconstruction via Online Restoration (CVPR 2025 · better on 6 of 6 shared results)
- StreetCrafter: Street View Synthesis with Controllable Video Diffusion Models (CVPR 2025 · better on 4 of 6 shared results)
- S3R-GS: Streamlining the Pipeline for Large-Scale Street Scene Reconstruction (ICCV 2025 · better on 4 of 4 shared results)
- RGE-GS: Reward-Guided Expansive Driving Scene Reconstruction via Diffusion Priors (ICCV 2025 · better on 2 of 2 shared results)
3 more not listed.
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.
- EmerNeRF — EmerNeRF: Emergent Spatial-Temporal Scene Decomposition via Self-Supervision (ICLR 2024) · 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
- DeSiRe-GS: 4D Street Gaussians for Static-Dynamic Decomposition and Surface Reconstruction for Urban Driving Scenes (CVPR 2025 · similarity 0.92)
- HUGS: Holistic Urban 3D Scene Understanding via Gaussian Splatting (CVPR 2024 · similarity 0.89)
- DrivingGaussian: Composite Gaussian Splatting for Surrounding Dynamic Autonomous Driving Scenes (CVPR 2024 · similarity 0.89)
- 3D StreetUnveiler with Semantic-aware 2DGS - a simple baseline (ICLR 2025 · similarity 0.89)
- EVolSplat: Efficient Volume-based Gaussian Splatting for Urban View Synthesis (CVPR 2025 · similarity 0.89)
- Dynamic 3D Gaussian Fields for Urban Areas (NeurIPS 2024 · similarity 0.89)
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.
