Research map / Point clouds / Shape and surface completion
Structure-to-Intensity Diffusion for Adverse-Weather LiDAR Generation
Peiyang Ni, Longyu Yang, Lu Zhang, Kuniaki Saito, Yap-Peng Tan, Fumin Shen, Heng Tao Shen, Xiaofeng Zhu, Ping Hu
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 WeatherGen MDP simulation with RPWS (ablation) | WeatherGen (SMG backbone) · STF snow, no fine-tuning | FPD: 271.12 -> 109.42 | Improved |
| replace SMG backbone with SiD (ablation) | RPWS-trained generator · STF snow, no fine-tuning | FPD: 109.42 -> 91.28 | Improved |
| add fine-tuning on real STF data (ablation) | RPWS + SiD · STF snow | FPD: 91.28 -> 32.38 | Improved |
| add reflectance degradation to RPWS (ablation) | RPWS + SiD · STF snow | FMD: 8.25 -> 3.67 | Improved |
| add geometry degradation to RPWS (ablation) | RPWS + SiD · STF snow | FPD: 475.11 -> 91.28 | Improved |
| condition intensity denoising on structure (S->I) versus independent (ablation) | SiD · STF snow | FMD: 4.92 -> 2.28 | Improved |
| reverse conditioning direction to I->S (ablation) | SiD · STF snow | FPD: 32.38 -> 66.20 | Hurt |
| use SiD instead of WeatherGen with only 10% real adverse data | SiD · STF snow, varying real data ratio | FMD | Improved |
| use SiD on clear-weather generation | SiD · KITTI-360 | FMD: 1.86 -> 1.75 FMD; FPD 5.62 vs 4.81 for Text2LiDAR | Mixed |
| use SiD (two-branch) instead of WeatherGen in parameters and compute | SiD · efficiency | parameters / GFLOPs: 38.06M and 1.599 GFLOPs vs 32.59M and 1.398 GFLOPs | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- WeatherGen — WeatherGen: A Unified Diverse Weather Generator for LiDAR Point Clouds via Spider Mamba Diffusion (CVPR 2025) · better on 12 of 12 shared results
- Text2LiDAR — Text2LiDAR: Text-guided LiDAR Point Clouds Generation via Equirectangular Transformer (ECCV 2024) · 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
- WeatherGen: A Unified Diverse Weather Generator for LiDAR Point Clouds via Spider Mamba Diffusion (CVPR 2025 · similarity 0.90)
- Towards Explicit Geometry-Reflectance Collaboration for Generalized LiDAR Segmentation in Adverse Weather (CVPR 2025 · similarity 0.86)
- Scaling Diffusion Models to Real-World 3D LiDAR Scene Completion (CVPR 2024 · similarity 0.86)
- Towards Realistic Scene Generation with LiDAR Diffusion Models (CVPR 2024 · similarity 0.85)
- UAWTrack: Universal 3D Single Object Tracking in Adverse Weather (AAAI 2025 · similarity 0.85)
- GS-LiDAR: Generating Realistic LiDAR Point Clouds with Panoramic Gaussian Splatting (ICLR 2025 · similarity 0.85)
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.
