Research map / Point clouds / Shape and surface completion
RadarGen: Automotive Radar Point Cloud Generation from Cameras
Tomer Borreda, Fangqiang Ding, Sanja Fidler, Shengyu Huang, Or Litany
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 feedforward RGB2Point baseline with RadarGen diffusion | radar generation · MAN TruckScenes Foreground | Hit Rate: 0.37 -> 0.66 | Improved |
| replace baseline with RadarGen | radar generation · MAN TruckScenes Entire Area | MMD Loc.: 0.368 ± 0.151 -> 0.056 ± 0.062 | Improved |
| replace baseline with RadarGen | radar generation · MAN TruckScenes Entire Area | CD Full: 0.038 ± 0.009 -> 0.040 ± 0.008 | Hurt |
| replace baseline with RadarGen | radar generation · MAN TruckScenes Entire Area | DA F1: 0.14 ± 0.09 -> 0.24 ± 0.12 | Improved |
| use generated radar instead of baseline radar for detector | VoxelNeXt detector · MAN TruckScenes | NDS: 0.30 vs nearly zero (real data 0.48) | Improved |
| add semantic map to BEV conditioning (ablation) | RadarGen · MAN TruckScenes Entire Area | MMD RCS: 0.12 ± 0.26 -> 0.09 ± 0.15 | Improved |
| add appearance map to BEV conditioning (ablation) | RadarGen · MAN TruckScenes Entire Area | MMD Doppler: 0.35 ± 0.86 -> 0.31 ± 0.74 | Improved |
| add velocity map to BEV conditioning (ablation) | RadarGen · MAN TruckScenes Entire Area | MMD Doppler: 0.34 ± 0.80 -> 0.31 ± 0.74 | Improved |
| add semantic map to BEV conditioning (ablation) | RadarGen · MAN TruckScenes Entire Area | CD Loc.: 1.72 ± 0.40 -> 1.68 ± 0.39 | Improved |
| replace multi-view camera conditioning with BEV conditioning (ablation) | RadarGen · MAN TruckScenes Entire Area | CD Loc.: 1.88 ± 0.50 -> 1.68 ± 0.39 | Improved |
| replace multi-view camera conditioning with BEV conditioning (ablation) | RadarGen · MAN TruckScenes Entire Area | MMD Doppler: 0.24 ± 0.85 -> 0.31 ± 0.74 | Hurt |
| replace multi-view conditioning with BEV conditioning (ablation) | RadarGen · MAN TruckScenes | training time: 9 days -> 2 days | Improved |
Closest papers by meaning
- RaLD: Generating High-Resolution 3D Radar Point Clouds with Latent Diffusion (AAAI 2026 · similarity 0.88)
- Effective Multi-sensor Conditioning for Street-view Novel-view Synthesis (NeurIPS 2026 · similarity 0.86)
- Selection, Not Fusion: Radar-Modulated State Space Models for Radar-Camera Depth Estimation (NeurIPS 2026 · similarity 0.85)
- FreeSim: Toward Free-viewpoint Camera Simulation in Driving Scenes (CVPR 2025 · similarity 0.84)
- Visual Point Cloud Forecasting enables Scalable Autonomous Driving (CVPR 2024 · similarity 0.84)
- StreetCrafter: Street View Synthesis with Controllable Video Diffusion Models (CVPR 2025 · similarity 0.84)
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.
