Research map / Point clouds / Segmentation and detection
Structure Aware Single-Stage 3D Object Detection From Point Cloud
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 point-level foreground segmentation supervision (ablation) | SA-SSD · KITTI val, Car | 3D AP: 82.88 -> 83.36; 92.10 -> 92.53; 79.96 -> 80.13 | Helped |
| add point-wise center estimation supervision (ablation) | SA-SSD with segmentation · KITTI val, Car | 3D AP: 83.36 -> 83.93; 92.53 -> 92.86; 80.13 -> 80.91 | Helped |
| add PSWarp confidence alignment (ablation) | SA-SSD with segmentation and center estimation · KITTI val, Car | 3D AP: 83.93 -> 84.30; 92.86 -> 93.23; 80.91 -> 81.36 | Helped |
| use point-level segmentation and center estimation auxiliary tasks | SECOND-based detector · KITTI test, Car | 3D AP: SECOND baseline: 83.34, 72.55, 65.82; SA-SSD: 88.75, 79.79, 74.16 | Helped |
| align classification confidence to predicted boxes with PSWarp (ablation) | SA-SSD · KITTI val, Car moderate | 3D AP: 84.30 vs 84.35 | No clear effect |
| replace PSWarp with PSRoIAlign (ablation) | SA-SSD · KITTI val, Car moderate | runtime: ∼0.4ms vs ∼4ms | Hurt |
| increase PSWarp alignment resolution from 2 × 3 to 4 × 7 (ablation) | SA-SSD · KITTI val, Car moderate | 3D AP: 84.05 -> 84.30 | Helped |
| increase PSRoIAlign alignment resolution from 2 × 3 to 4 × 7 (ablation) | SA-SSD · KITTI val, Car moderate | 3D AP: 84.18 -> 84.35 | Helped |
Closest papers by meaning
- VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection (CVPR 2018 · similarity 0.89)
- PointPillars: Fast Encoders for Object Detection From Point Clouds (CVPR 2019 · similarity 0.89)
- Not All Points Are Equal: Learning Highly Efficient Point-Based Detectors for 3D LiDAR Point Clouds (CVPR 2022 · similarity 0.89)
- Multi-View 3D Object Detection Network for Autonomous Driving (CVPR 2017 · similarity 0.89)
- HVPR: Hybrid Voxel-Point Representation for Single-Stage 3D Object Detection (CVPR 2021 · similarity 0.89)
- Leveraging Imagery Data with Spatial Point Prior for Weakly Semi-supervised 3D Object Detection (AAAI 2024 · 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.