Research map / Detection and segmentation / 3D detection and driving
RAF: Reliability-Aware Fusion of Camera, LiDAR, and 4D RADAR for Robust 3D Object Detection in Adverse Weather
Heejun Park, Jaeseok Jeong, KUK-JIN YOON
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 camera branch with reliability supervision (RAF) | L4DR · K-Radar Total | AP_BEV: 77.5 -> 82.0 | Improved |
| add camera branch with reliability supervision (RAF) | 3D-LRF · K-Radar Total | AP_BEV / AP_3D: 64.5 -> 71.0 / 35.8 -> 43.2 | Improved |
| add unsupervised confidence gating (detection loss only) (ablation) | L4DR with pretrained frozen backbone · K-Radar Total | AP_BEV: 78.6 -> 76.4 | Hurt |
| add unsupervised confidence gating (detection loss only) (ablation) | 3D-LRF with frozen backbone · K-Radar Total | AP_3D: 39.1 -> 33.1 | Hurt |
| add weakly-supervised reliability estimation (sim + rel losses) (ablation) | L4DR · K-Radar Total | AP_BEV: 76.4 -> 80.2 | Improved |
| add Calibration-Aware Local Matching (CALM) (ablation) | L4DR · K-Radar Total | AP_BEV / AP_3D: 80.2 -> 82.0 / 55.4 -> 57.4 | Improved |
| add CALM (ablation) | 3D-LRF · K-Radar Total | AP_BEV / AP_3D: 70.5 -> 71.0 / 41.9 -> 43.2 | Improved |
| use pretrained frozen LiDAR-RADAR backbone instead of joint training (ablation) | L4DR / 3D-LRF · K-Radar Total | AP_BEV: 77.8 -> 78.6 (L4DR); 65.7 -> 68.3 (3D-LRF) | Improved |
| enlarge CALM window from 5 to 7 (ablation) | RAF · K-Radar Total | AP_3D: 57.4 -> 55.4 (L4DR); 43.2 -> 39.8 (3D-LRF) | Hurt |
| use hard gating alpha=0 instead of alpha=0.2 (ablation) | RAF · K-Radar Total | AP_3D: 3D-LRF 43.2 -> 45.0; L4DR 57.4 -> 58.5 | Mixed |
| replace CNN reliability estimator with ViT (ablation) | RAF · K-Radar Total | AP_3D: 57.4 -> 56.4 (L4DR); 43.2 -> 40.2 (3D-LRF) | Hurt |
| naive camera fusion in adverse weather (Joint/Frozen) vs LiDAR-RADAR baseline | L4DR / 3D-LRF · K-Radar Noisy split | AP_BEV: L4DR 59.0 -> 55.6; 3D-LRF 38.8 -> 25.4 (Joint) | Hurt |
Closest papers by meaning
- L4DR: LiDAR-4DRadar Fusion for Weather-Robust 3D Object Detection (AAAI 2025 · similarity 0.92)
- Towards Robust 3D Object Detection with LiDAR and 4D Radar Fusion in Various Weather Conditions (CVPR 2024 · similarity 0.91)
- RobuRCDet: Enhancing Robustness of Radar-Camera Fusion in Bird's Eye View for 3D Object Detection (ICLR 2025 · similarity 0.90)
- An Empirical Study of the Generalization Ability of Lidar 3D Object Detectors to Unseen Domains (CVPR 2024 · similarity 0.89)
- Sunshine to Rainstorm: Cross-Weather Knowledge Distillation for Robust 3D Object Detection (AAAI 2024 · similarity 0.88)
- Availability-aware Sensor Fusion via Unified Canonical Space (NeurIPS 2025 · similarity 0.88)
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.
