Research map / Point clouds / Segmentation and detection
Large-Scale Point Cloud Semantic Segmentation With Superpoint Graphs
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 |
|---|---|---|---|
| Use SPG contextual segmentation instead of the strongest reported baseline | Semantic3D reduced test set · Semantic3D reduced test set | mIoU: 54.2 -> 73.2 | Helped |
| Use SPG contextual segmentation instead of the strongest reported baseline | Semantic3D full test set · Semantic3D full test set | mIoU: +8.8 mIoU points | Helped |
| Use SPG instead of SEGCloud | S3DIS Area 5 · S3DIS Area 5 | mIoU: 48.92 -> 58.04 | Helped |
| Use SPG instead of Engelmann et al. | S3DIS 6-fold cross validation · S3DIS 6-fold cross validation | mIoU: 49.7 -> 62.1 | Helped |
| Add contextual segmentation with GRUs and edge-conditioned messages | Unary PointNet13 · S3DIS 6-fold cross validation | mIoU: 40.0 -> 62.1 | Helped |
| Add end-to-end CRF-ECC message passing | PointNet13 unary predictions · S3DIS 6-fold cross validation | mIoU: 40.0 -> 55.3 | Helped |
| Add scalar-transition CRF post-processing | Unary · S3DIS 6-fold cross validation | mIoU: 40.0 -> 40.7 | Helped |
| Use 32-dimensional rather than 13-dimensional embeddings | GRU contextual segmentation · S3DIS 6-fold cross validation | mIoU: 58.5 -> 62.1 | Helped |
| Add input gating to GRU (ablation) | Best · S3DIS 6-fold cross validation | mIoU: 57.5 -> 62.1 | Helped |
| Concatenate hidden states from all GRU steps for output (ablation) | Best · S3DIS 6-fold cross validation | mIoU: 57.7 -> 62.1 | Helped |
| Use informative superedge features rather than a constant scalar (ablation) | Best · S3DIS 6-fold cross validation | mIoU: 39.9 -> 62.1 | Helped |
| Use the proposed lightweight ECC-VV formulation instead of ECC-MV (ablation) | Best · S3DIS 6-fold cross validation | mIoU: 62.1 -> 59.4 | Hurt |
Closest papers by meaning
- Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation (CVPR 2021 · similarity 0.91)
- SSPC-Net: Semi-supervised Semantic 3D Point Cloud Segmentation Network (AAAI 2021 · similarity 0.90)
- LogoSP: Local-global Grouping of Superpoints for Unsupervised Semantic Segmentation of 3D Point Clouds (CVPR 2025 · similarity 0.89)
- Geometrically-driven Aggregation for Zero-shot 3D Point Cloud Understanding (CVPR 2024 · similarity 0.88)
- LiDAR-Net: A Real-scanned 3D Point Cloud Dataset for Indoor Scenes (CVPR 2024 · similarity 0.88)
- Spherical Frustum Sparse Convolution Network for LiDAR Point Cloud Semantic Segmentation (NeurIPS 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.