Research map / Detection and segmentation / 3D detection and driving
RaGS: Unleashing 3D Gaussian Splatting from 4D Radar and Monocular Cue for 3D Object Detection
Xiaokai Bai, Chenxu Zhou, Lianqing Zheng, Jianan Liu, Si-Yuan Cao, Xiaohan Zhang, Yiming Li, Zhengzhuang Zhang, Hui-Liang Shen
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 Frustum-based Localization Initiation (FLI) to the baseline (ablation) | RaGS baseline · View-of-Delft val (half epochs) | mAP entire area: 55.33 -> 57.40 | Improved |
| add Iterative Multimodal Aggregation (IMA) on top of FLI (ablation) | RaGS · View-of-Delft val (half epochs) | mAP entire area: 57.40 -> 59.12 | Improved |
| add Multi-level Gaussian Fusion (MGF) on top of FLI and IMA (ablation) | RaGS · View-of-Delft val (half epochs) | mAP entire area: 59.12 -> 59.45 | Improved |
| add 4D radar points to FLI frustum initialization (ablation) | FLI · View-of-Delft val (half epochs) | mAP driving corridor: 73.10 -> 75.78 | Improved |
| add depth back-projection to FLI (ablation) | FLI · View-of-Delft val (half epochs) | mAP entire area: 56.72 -> 57.40 | Improved |
| add 3D deformable cross attention to IMA (ablation) | IMA · View-of-Delft val (half epochs) | mAP entire area: 57.40 -> 58.79 | Improved |
| add radar sparse pillars to IMA (ablation) | IMA · View-of-Delft val (half epochs) | mAP entire area: 58.79 -> 58.93 | Improved |
| add frustum-space position refinement to IMA (ablation) | IMA · View-of-Delft val (half epochs) | mAP entire area: 58.93 -> 59.45 | Improved |
| increase Gaussian anchors from 12800 to 19200 (ablation) | RaGS · View-of-Delft val | mAP entire area: 59.45 -> 59.47 | No clear effect |
| increase Gaussian anchors from 3200 to 12800 (ablation) | RaGS · View-of-Delft val | mAP entire area: 54.77 -> 59.45 | Improved |
| replace BEV-grid fusion with Gaussian-based fusion (RaGS vs SGDet3D) | 4D radar-camera fusion · View-of-Delft val | GFLOPs: 1309.3 -> 781.0 | Improved |
| use RaGS instead of LXL under calibration disturbance of +-5 deg and +-0.5m | RaGS · View-of-Delft val | mAP entire area: RaGS 56.66 vs LXL 50.25 | Improved |
Later papers that compared with this method
1 later accepted paper reports RaGS as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- RPGFusion: 4D Radar Prior-Guided Multi-Modal Fusion for 3D Detection (CVPR 2026 · better on 2 of 2 shared results)
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.
- RaCFormer — RaCFormer: Towards High-Quality 3D Object Detection via Query-based Radar-Camera Fusion (CVPR 2025) · better on 1 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
- CRT-Fusion: Camera, Radar, Temporal Fusion Using Motion Information for 3D Object Detection (NeurIPS 2024 · similarity 0.87)
- HGSFusion: Radar-Camera Fusion with Hybrid Generation and Synchronization for 3D Object Detection (AAAI 2025 · similarity 0.86)
- RaCFormer: Towards High-Quality 3D Object Detection via Query-based Radar-Camera Fusion (CVPR 2025 · similarity 0.86)
- RobuRCDet: Enhancing Robustness of Radar-Camera Fusion in Bird's Eye View for 3D Object Detection (ICLR 2025 · similarity 0.86)
- Gaussian-Det: Learning Closed-Surface Gaussians for 3D Object Detection (ICLR 2025 · similarity 0.85)
- IS-Fusion: Instance-Scene Collaborative Fusion for Multimodal 3D Object Detection (CVPR 2024 · 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.
