Research map / 3D vision / Gaussian splatting
LiDAR-GS++: Improving LiDAR Gaussian Reconstruction via Diffusion Priors
Qifeng Chen, Jiarun Liu, Rengan Xie, Tao Tang, Sicong Du, Yiru Zhao, Yuchi Huo, Sheng Yang
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 |
|---|---|---|---|
| replace vanilla 2DGS with neural 2DGS field (ablation) | LiDAR-GS++ · Para-Lane extrapolation @3.5m | Chamfer Distance: 0.417 -> 0.102 | Improved |
| add diffusion-generated extrapolated scans as supervision (ablation) | LiDAR-GS++ · Para-Lane extrapolation @3.5m | Chamfer Distance: 0.264 -> 0.102 | Improved |
| add diffusion-generated supervision (ablation) | LiDAR-GS++ · Para-Lane interpolation | Chamfer Distance: 0.079 -> 0.079 | No clear effect |
| add depth distortion-aware distillation (DDAD) instead of fully injecting generated data (ablation) | LiDAR-GS++ · Para-Lane extrapolation @3.5m | Chamfer Distance: 0.163 -> 0.102 | Improved |
| add DDAD (ablation) | LiDAR-GS++ · Para-Lane interpolation | PSNR: 32.013 -> 32.232 | Improved |
| add diffusion prior and DDAD to GS-LiDAR (ablation) | GS-LiDAR · Para-Lane extrapolation @3.5m | Chamfer Distance: 0.305 -> 0.116 | Improved |
| add Fourier position encoding to the generator input (ablation) | controllable LiDAR generation · Waymo extrapolation | FRID: 33.62 -> 28.39 | Improved |
| add wavelet transforms in up/down-sampling (ablation) | controllable LiDAR generation · Waymo extrapolation | FRID: 29.31 -> 28.39 | Improved |
| condition on coarse rendered LiDAR instead of bounding boxes and map (ablation) | controllable LiDAR generation · Waymo extrapolation | FRID: 46.74 -> 28.39 | Improved |
| use LiDAR-GS++ instead of LiDAR-GS on extrapolated views | Para-Lane extrapolation · Para-Lane extrapolation @3.5m | Chamfer Distance: 0.270 -> 0.102 | Improved |
Later papers that compared with this method
1 later accepted paper reports LiDAR-GS as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- LiDAR-EVS: Enhance Extrapolated View Synthesis for 3D Gaussian Splatting with Pseudo-LiDAR Supervision (ECCV 2026 · better on 1 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- LiDAR-RT — LiDAR-RT: Gaussian-based Ray Tracing for Dynamic LiDAR Re-simulation (CVPR 2025) · better on 0 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
- GS-LiDAR: Generating Realistic LiDAR Point Clouds with Panoramic Gaussian Splatting (ICLR 2025 · similarity 0.92)
- D$^2$GS: Dense Depth Regularization for LiDAR-free Urban Scene Reconstruction (NeurIPS 2025 · similarity 0.92)
- LiDAR-RT: Gaussian-based Ray Tracing for Dynamic LiDAR Re-simulation (CVPR 2025 · similarity 0.89)
- SplatAD: Real-Time Lidar and Camera Rendering with 3D Gaussian Splatting for Autonomous Driving (CVPR 2025 · similarity 0.89)
- Horizon-GS: Unified 3D Gaussian Splatting for Large-Scale Aerial-to-Ground Scenes (CVPR 2025 · similarity 0.88)
- Wild-GS: Real-Time Novel View Synthesis from Unconstrained Photo Collections (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.
