Research map / Point clouds / Segmentation and detection
Sparse Fuse Dense: Towards High Quality 3D Detection With Depth Completion
Xiaopei Wu, Liang Peng, Honghui Yang, Liang Xie, Chenxi Huang, Chengqi Deng, Haifeng Liu, Deng Cai
Check 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 pseudo-cloud RoI features and 3D-GAF to the Voxel-RCNN baseline (ablation) | Voxel-RCNN · KITTI val, car 3D R40, moderate | AP: 85.47 -> 86.57 | Improved |
| Add CPConv for pseudo-cloud feature extraction (ablation) | Voxel-RCNN with 3D-GAF · KITTI val, car 3D R40, moderate | AP: +1.99% AP | Improved |
| Enable SynAugment and LiDAR-style data augmentation (ablation) | SFD baseline experiment (a) · KITTI val, car 3D R40, moderate | AP: 78.49 -> 85.47 | Improved |
| Enable SynAugment and LiDAR-style data augmentation (ablation) | SFD experiment (b) with 3D-GAF · KITTI val, car 3D R40, moderate | AP: 80.31 -> 86.57 | Improved |
| Use 3D RoIs instead of 2D RoIs for fusion (ablation) | SFD · KITTI val, car 3D R40, moderate | AP: 85.27 -> 87.77 | Improved |
| Add grid-wise fusion (ablation) | SFD with 3D RoI fusion · KITTI val, car 3D R40, moderate | AP: 87.77 -> 88.23 | Improved |
| Add attentive weighting to grid-wise fusion (ablation) | SFD with 3D grid-wise fusion · KITTI val, car 3D R40, moderate | AP: 88.23 -> 88.56 | Improved |
| Add SFD to the LiDAR detector | PointRCNN · KITTI val, car 3D R40 | AP: +3.39 | Improved |
| Add SFD to the LiDAR detector | Part-A² · KITTI val, car 3D R40 | AP: +3.17 | Improved |
| Add SFD to the LiDAR detector | SECOND · KITTI val, car 3D R40 | AP: +5.44 | Improved |
| Add SFD to Voxel-RCNN | Voxel-RCNN · KITTI val, car 3D R40, moderate, 40m-Inf | AP: +6.88 | Improved |
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.
- SE-SSD — SE-SSD: Self-Ensembling Single-Stage Object Detector From Point Cloud (CVPR 2021) · better on 3 of 3 shared results
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
- Fully Sparse Fusion for 3D Object Detection (TPAMI 2024 · similarity 0.91)
- CCF: Complementary Collaborative Fusion for Domain Generalized Multi-Modal 3D Object Detection (CVPR 2026 · similarity 0.91)
- SparseLIF: High-Performance Sparse LiDAR-Camera Fusion for 3D Object Detection (ECCV 2024 · similarity 0.91)
- Voxel or Pillar: Exploring Efficient Point Cloud Representation for 3D Object Detection (AAAI 2024 · similarity 0.90)
- Beyond Virtual Points: Depth-Enhanced LiDAR-only 3D Object Detection with Semi-Supervised Learning (Student Abstract) (AAAI 2025 · similarity 0.88)
- GAFusion: Adaptive Fusing LiDAR and Camera with Multiple Guidance for 3D Object Detection (CVPR 2024 · 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.
