Research map / 3D vision / Gaussian splatting
AutoWeather4D: Autonomous Driving Video Weather Conversion via G-Buffer Dual-Pass Editing
Tianyu Liu, Weitao Xiong, Kunming Luo, Manyuan Zhang, Peng Li, Yuan Liu, Ping Tan
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 feed-forward 4D reconstruction backbone instead of integer-quantized depth from inverse rendering alone (ablation) | AutoWeather4D local relighting · Local illumination rendering (qualitative) | visual artifacts (aliasing) | Improved |
| augment HRDA training with AutoWeather4D videos vs none | HRDA · ACDC | mIoU: 49.20 -> 49.81 | Improved |
| augment HRDA training with AutoWeather4D videos vs none | HRDA · Dark Zurich | mIoU: 23.92 -> 24.09 | Improved |
| augment HRDA training with Cosmos videos vs none | HRDA · Dark Zurich | mIoU: 23.92 -> 23.93 | No clear effect |
| use AutoWeather4D instead of Cosmos for augmentation | HRDA · ACDC | mAcc: 62.31 -> 62.52 | Improved |
| use AutoWeather4D instead of Cosmos for augmentation | HRDA · Dark Zurich | mAcc: 39.52 -> 39.73 | Improved |
| AutoWeather4D versus Cosmos-Transfer2.5 on editing adherence | weather conversion · Waymo weather conversion | CLIP score: 0.2558 -> 0.2586 | No clear effect |
| AutoWeather4D versus Cosmos-Transfer2.5 on identity stability | weather conversion · Waymo weather conversion | Vehicle CLIP cosine similarity: 0.837 -> 0.871 | Improved |
| AutoWeather4D versus WAN-FUN 2.2 on human preference | weather conversion · Waymo weather conversion | Human evaluation: 0.668 -> 0.826 | Improved |
| AutoWeather4D versus Cosmos-Transfer2.5 on structural consistency | weather conversion · Waymo weather conversion | Vehicle 3D detection IoU: 0.913 -> 0.915 | No clear effect |
Closest papers by meaning
- IntrinsicWeather: Controllable Weather Editing in Intrinsic Space (CVPR 2026 · similarity 0.89)
- SceneCrafter: Controllable Multi-View Driving Scene Editing (CVPR 2025 · similarity 0.87)
- UrbanCAD: Towards Highly Controllable and Photorealistic 3D Vehicles for Urban Scene Simulation (CVPR 2025 · similarity 0.85)
- DriveDreamer4D: World Models Are Effective Data Machines for 4D Driving Scene Representation (CVPR 2025 · similarity 0.85)
- X-Scene: Large-Scale Driving Scene Generation with High Fidelity and Flexible Controllability (NeurIPS 2025 · similarity 0.85)
- DrivingSphere: Building a High-fidelity 4D World for Closed-loop Simulation (CVPR 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.
