Research map / 3D vision / Gaussian splatting
BEV-GS: Feed-forward Gaussian Splatting in Bird’s-Eye-View for Road Reconstruction
Wenhua Wu, Tong Zhao, Chensheng Peng, Lei Yang, Zhe Liu, 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 |
|---|---|---|---|
| Use separate geometry and texture backbones (ablation) | BEV-GS · RSRD ablation | AAE (cm): 1.89 -> 1.73 | Improved |
| Use separate texture backbone (ablation) | BEV-GS · RSRD ablation | PSNR (dB): 24.90 -> 28.36 | Improved |
| Add elevation guidance to texture feature query (ablation) | BEV-GS · RSRD ablation | PSNR (dB): 25.00 -> 28.36 | Improved |
| Add elevation guidance to texture feature query (ablation) | BEV-GS · RSRD ablation | AAE (cm): 1.73 -> 1.73 | No clear effect |
| Upsample Gaussian grid to 2x texture grid (ablation) | BEV-GS · RSRD grid resolution ablation | PSNR (dB): 27.63 -> 28.36 | Improved |
| Increase texture grid resolution (ablation) | BEV-GS · RSRD grid resolution ablation | PSNR (dB): 25.46 -> 27.63 | Improved |
| Add test-time optimization (90 iterations) (ablation) | BEV-GS · RSRD test-time optimization | PSNR (dB): 28.361 -> 32.800 | Improved |
| BEV elevation representation vs PV depth estimation | BEV-GS · RSRD (road elevation) | AAE (cm): 2.30 -> 1.73 | Improved |
| BEV grid Gaussians vs per-pixel Gaussians | BEV-GS · RSRD (novel view synthesis) | PSNR (dB): 23.27 -> 28.36 | Improved |
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.
- Splatter Image — Splatter Image: Ultra-Fast Single-View 3D Reconstruction (CVPR 2024) · better on 5 of 5 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
- Horizon-GS: Unified 3D Gaussian Splatting for Large-Scale Aerial-to-Ground Scenes (CVPR 2025 · similarity 0.91)
- DeSiRe-GS: 4D Street Gaussians for Static-Dynamic Decomposition and Surface Reconstruction for Urban Driving Scenes (CVPR 2025 · similarity 0.90)
- EVolSplat: Efficient Volume-based Gaussian Splatting for Urban View Synthesis (CVPR 2025 · similarity 0.89)
- Wild-GS: Real-Time Novel View Synthesis from Unconstrained Photo Collections (NeurIPS 2024 · similarity 0.88)
- D$^2$GS: Dense Depth Regularization for LiDAR-free Urban Scene Reconstruction (NeurIPS 2025 · similarity 0.88)
- CityGaussianV2: Efficient and Geometrically Accurate Reconstruction for Large-Scale Scenes (ICLR 2025 · 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.
