Research map / Detection and segmentation / 3D detection and driving
Refine3D: Scene-Adaptive Reference Point Refinement for Sparse 3D Object Detection
Fan Li, Jing Lu, Yunlu Xu, Changhong Wu, Tao Xu, Zhaoyi Xiang, Yi Niu
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 SAR-Head offset MLP (no embedding branch) to RayDN (ablation) | RayDN · nuScenes val (ResNet50, 24 ep) | mAP: 0.409 -> 0.442 | Improved |
| add position embedding branch (MLPembed) to SAR-Head (ablation) | RayDN + MLPoffset · nuScenes val (ResNet50, 24 ep) | NDS: 0.532 -> 0.548 | Improved |
| add RPD-Loss on top of SAR-Head (ablation) | RayDN + SAR-Head · nuScenes val (ResNet50, 24 ep) | mAP: 0.447 -> 0.455 | Improved |
| add RPD-Loss on top of SAR-Head (ablation) | RayDN + SAR-Head · nuScenes val (ResNet50, 24 ep) | mATE: 0.589 -> 0.595 | Hurt |
| add Refine3D to RayDN | RayDN · nuScenes val full set (ResNet50) | mAP: +/-0.0% | No clear effect |
| add Refine3D to RayDN | RayDN · nuScenes val imbalanced subset (ResNet50) | mAP: +2.4% | Improved |
| add Refine3D to RayDN | RayDN · Argoverse 2 val | mAP: 0.223 -> 0.259 | Improved |
| add Refine3D to RayDN | RayDN · nuScenes test (V2-99) | NDS: 0.645 -> 0.658 | Improved |
| add Refine3D to PETR | PETR · nuScenes val | mAP: 0.370 -> 0.387 | Improved |
| add Refine3D to StreamPETR | StreamPETR · nuScenes val | NDS: 0.490 -> 0.530 | Improved |
| add Refine3D to Sparse4Dv3, which already refines reference points | Sparse4Dv3 · nuScenes val | mAP: 0.397 -> 0.442 | Improved |
| add Refine3D to RayDN (ResNet101, 512x1408) | RayDN · nuScenes val | NDS: 0.604 -> 0.624 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Far3D — Far3D: Expanding the Horizon for Surround-View 3D Object Detection (AAAI 2024) · better on 1 of 1 shared result
- PETR — PETR: Position Embedding Transformation for Multi-View 3D Object Detection (ECCV 2022) · 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
- DiffuBox: Refining 3D Object Detection with Point Diffusion (NeurIPS 2024 · similarity 0.88)
- RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection (NeurIPS 2026 · similarity 0.87)
- Ada3D : Exploiting the Spatial Redundancy with Adaptive Inference for Efficient 3D Object Detection (ICCV 2023 · similarity 0.87)
- Sparse Point Guided 3D Lane Detection (ICCV 2023 · similarity 0.86)
- SAFDNet: A Simple and Effective Network for Fully Sparse 3D Object Detection (CVPR 2024 · similarity 0.86)
- Leveraging Imagery Data with Spatial Point Prior for Weakly Semi-supervised 3D Object Detection (AAAI 2024 · similarity 0.86)
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.
