Research map / Point clouds / Segmentation and detection
SASA: Semantics-Augmented Set Abstraction for Point-Based 3D Object Detection
Read the paperCheck 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 SASA to 3DSSD | 3DSSD · KITTI test, Car | AP: 88.36 -> 88.76; 79.57 -> 82.16; 74.55 -> 77.16 | Helped |
| Add SASA to 3DSSD | 3DSSD · nuScenes validation | NDS: 56.4 -> 61.0 | Helped |
| Add SASA to 3DSSD | 3DSSD · nuScenes validation | mAP: 42.6 -> 45.0 | Helped |
| Add SASA to PointRCNN | PointRCNN · KITTI val, Car | AP: 91.57 -> 92.13; 82.24 -> 82.64; 80.45 -> 82.40 | Helped |
| Add point segmentation while retaining FPS and no fusion sampling (ablation) | 3DSSD · KITTI val, Car | AP: 82.75 -> 82.83 | No clear effect |
| Replace F-FPS with S-FPS at γ=1 (ablation) | 3DSSD · KITTI val, Car Moderate | AP: 83.46 -> 85.76 | Helped |
| Use S-FPS with γ=0.1 (ablation) | 3DSSD · KITTI val, Car Moderate | AP: 83.46 -> 83.16 | Hurt |
| Use S-FPS with γ=10 (ablation) | 3DSSD · KITTI val, Car Moderate | AP: 83.46 -> 83.41 | No clear effect |
| Use S-FPS with γ=100 (ablation) | 3DSSD · KITTI val, Car Moderate | AP: 83.46 -> 82.35 | Hurt |
| Use S-FPS with γ=100 (ablation) | 3DSSD · KITTI val, Car | point recall: 92.10 -> 91.19 | Hurt |
| Add SASA to 3DSSD | 3DSSD · KITTI val, Car | mAP: 85.73 -> 87.02 | Helped |
Closest papers by meaning
- Voxel Set Transformer: A Set-to-Set Approach to 3D Object Detection From Point Clouds (CVPR 2022 · similarity 0.88)
- PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detection (CVPR 2020 · similarity 0.86)
- 3DSSD: Point-Based 3D Single Stage Object Detector (CVPR 2020 · similarity 0.86)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.86)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · similarity 0.85)
- Point-Cache: Test-time Dynamic and Hierarchical Cache for Robust and Generalizable Point Cloud Analysis (CVPR 2025 · similarity 0.84)
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.