Research map / Point clouds / Segmentation and detection
A Hierarchical Graph Network for 3D Object Detection on Point Clouds
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 |
|---|---|---|---|
| Use HGNet instead of VoteNet | 3D object detection · SUN RGB-D V1 | mAP@0.25: 57.7 -> 61.6 | Helped |
| Use HGNet instead of VoteNet | 3D object detection · ScanNet-V2 | mAP@0.25: 58.6 -> 61.3 | Helped |
| Use HGNet instead of VoteNet | 3D object detection · ScanNet-V2 | mAP@0.50: 33.5 -> 34.4 | Helped |
| Add SA-GConv in place of simple GConv without position modelling (ablation) | HGNet · SUN RGB-D | mAP: ∼2% | Helped |
| Add the feature pyramid to combine multi-level features (ablation) | HGNet · SUN RGB-D | mAP: ∼2% | Helped |
| Add the ProRe Module to reason over proposals (ablation) | HGNet · SUN RGB-D | mAP: ∼2% | Helped |
| Use SA-DeGConv instead of arithmetic interpolation for up-sampling (ablation) | HGNet · SUN RGB-D | mAP: 0.4% | Helped |
| Use SA-GConv instead of PointNet++ set abstraction for voting features (ablation) | HGNet · SUN RGB-D | precise votes (%): 36.4 -> 42.5 | Helped |
| Add the ProRe Module to propagate proposal features (ablation) | HGNet · SUN RGB-D objects with very few points | AR: over 12% | Helped |
Closest papers by meaning
- Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud (CVPR 2020 · similarity 0.89)
- PC-RGNN: Point Cloud Completion and Graph Neural Network for 3D Object Detection (AAAI 2021 · similarity 0.88)
- Convolution in the Cloud: Learning Deformable Kernels in 3D Graph Convolution Networks for Point Cloud Analysis (CVPR 2020 · similarity 0.88)
- Object DGCNN: 3D Object Detection using Dynamic Graphs (NeurIPS 2021 · similarity 0.88)
- HGNet: Learning Hierarchical Geometry From Points, Edges, and Surfaces (CVPR 2023 · similarity 0.88)
- HEDNet: A Hierarchical Encoder-Decoder Network for 3D Object Detection in Point Clouds (NeurIPS 2023 · 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.