Research map / Detection and segmentation / 3D detection and driving
CRN: Camera Radar Net for Accurate, Robust, Efficient 3D Perception
Youngseok Kim, Juyeb Shin, Sanmin Kim, In-Jae Lee, Jun Won Choi, Dongsuk Kum
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 radar-assisted view transformation (depth plus radar occupancy) over depth-only (ablation) | CRN camera stream · nuScenes val | NDS: 43.9 -> 52.1 | Improved |
| replace depth distribution with raw radar occupancy for view transformation (ablation) | CRN camera stream · nuScenes val | NDS: 43.9 -> 33.6 | Hurt |
| use multi-modal deformable cross attention aggregation (MFA) instead of single-conv BEVFusion (ablation) | CRN · nuScenes val | NDS: 51.9 -> 53.4 | Improved |
| add two more fusion conv layers to BEVFusion (ablation) | BEVFusion · nuScenes val | NDS: 51.9 -> 51.9 | No clear effect |
| add RVT on top of MFA (ablation) | CRN · nuScenes val | NDS: 53.4 -> 53.9 | Improved |
| add sparse aggregation (top-k queries) (ablation) | CRN long-range · nuScenes val 102.4m Car mAP | Car mAP: mAP 56.9 -> 54.0, FPS 11.5 -> 14.0 | Mixed |
| keep only 1024 top-k queries instead of 4096 (ablation) | CRN sparse aggregation · nuScenes val 100m Car | AP: 54.0 -> 49.8 | Hurt |
| increase temporal BEV frames from 1 to 4 (ablation) | CRN · nuScenes val | NDS: 50.3 -> 56.0 | Improved |
| increase temporal frames from 4 to 5 (ablation) | CRN · nuScenes val | NDS: 56.0 -> 56.4 | No clear effect |
| fuse radar with camera using CRN under complete radar loss | CRN vs BEVFusion · nuScenes val, robustness, no radar | Car mAP: 34.4 -> 43.8 | Improved |
| add radar fusion to camera-only detector | BEVDepth · nuScenes val, night | mAP: 16.8 -> 30.4 | Improved |
Closest papers by meaning
- RobuRCDet: Enhancing Robustness of Radar-Camera Fusion in Bird's Eye View for 3D Object Detection (ICLR 2025 · similarity 0.91)
- RCBEVDet: Radar-camera Fusion in Bird's Eye View for 3D Object Detection (CVPR 2024 · similarity 0.91)
- CRT-Fusion: Camera, Radar, Temporal Fusion Using Motion Information for 3D Object Detection (NeurIPS 2024 · similarity 0.89)
- RaCFormer: Towards High-Quality 3D Object Detection via Query-based Radar-Camera Fusion (CVPR 2025 · similarity 0.89)
- CRKD: Enhanced Camera-Radar Object Detection with Cross-modality Knowledge Distillation (CVPR 2024 · similarity 0.88)
- HGSFusion: Radar-Camera Fusion with Hybrid Generation and Synchronization for 3D Object Detection (AAAI 2025 · similarity 0.87)
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.
