Research map / 3D vision / Gaussian splatting
EMD: Explicit Motion Modeling for High-Quality Street Gaussian Splatting
Xiaobao Wei, Qingpo Wuwu, Zhongyu Zhao, Zhuangzhe Wu, Nan Huang, Ming Lu, Ningning Ma, Shanghang Zhang
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 Gaussian embedding (ablation) | S3Gaussian + EMD · Waymo-D32 scene reconstruction | PSNR: 32.21 -> 32.50 | Improved |
| add temporal embedding (ablation) | S3Gaussian + EMD · Waymo-D32 scene reconstruction | PSNR: 32.23 -> 32.50 | Improved |
| add coarse deformation (ablation) | S3Gaussian + EMD · Waymo-D32 scene reconstruction | PSNR: 29.40 -> 32.50 | Improved |
| add fine deformation (ablation) | S3Gaussian + EMD · Waymo-D32 scene reconstruction | LPIPS: 0.118 -> 0.082 | Improved |
| increase Gaussian embedding dim from 32 to 128 (ablation) | S3Gaussian + EMD · Waymo-D32 | PSNR: 32.50 -> 32.22 | No clear effect |
| reduce Gaussian embedding dim from 32 to 8 (ablation) | S3Gaussian + EMD · Waymo-D32 | PSNR: 32.50 -> 31.25 | Hurt |
| reduce temporal embedding dim from 4 to 1 (ablation) | S3Gaussian + EMD · Waymo-D32 | PSNR: 32.50 -> 30.55 | Hurt |
| increase temporal embedding dim from 4 to 16 (ablation) | S3Gaussian + EMD · Waymo-D32 | PSNR: 32.50 -> 32.38 | No clear effect |
| add EMD to S3Gaussian | S3Gaussian · Waymo-D32 scene reconstruction | PSNR: 30.69 -> 32.50 | Improved |
| add EMD to S3Gaussian | S3Gaussian · Waymo-D32 novel view synthesis | PSNR: 26.62 -> 26.55 | No clear effect |
| add EMD to S3Gaussian | S3Gaussian · Waymo novel trajectory 1.0m | FID: 110.11 -> 70.26 | Improved |
| add EMD to DeSiRe-GS | DeSiRe-GS · Waymo Open scene reconstruction | LPIPS: 0.204 -> 0.130 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DeSiRe-GS — DeSiRe-GS: 4D Street Gaussians for Static-Dynamic Decomposition and Surface Reconstruction for Urban Driving Scenes (CVPR 2025) · better on 5 of 5 shared results
- OmniRe — OmniRe: Omni Urban Scene Reconstruction (ICLR 2025) · better on 2 of 2 shared results
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)
- SplatFlow: Self-Supervised Dynamic Gaussian Splatting in Neural Motion Flow Field for Autonomous Driving (CVPR 2025 · similarity 0.89)
- Motion Decoupled 3D Gaussian Splatting for Dynamic Object Representation (AAAI 2025 · similarity 0.88)
- DrivingGaussian: Composite Gaussian Splatting for Surrounding Dynamic Autonomous Driving Scenes (CVPR 2024 · similarity 0.87)
- 3D StreetUnveiler with Semantic-aware 2DGS - a simple baseline (ICLR 2025 · similarity 0.87)
- EGSRAL:An Enhanced 3D Gaussian Splatting Based Renderer with Automated Labeling for Large-Scale Driving Scene (AAAI 2025 · similarity 0.87)
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.
