Research map / Point clouds / Segmentation and detection
Point Cloud Oversegmentation With Graph-Structured Deep Metric Learning
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 |
|---|---|---|---|
| Replace unsupervised superpoint computation with SSP in SPG | SPG semantic segmentation on S3DIS · S3DIS 6-fold cross-validation | mIoU: 62.1 -> 68.4 | Helped |
| Replace unsupervised superpoint computation with SSP in SPG | SPG semantic segmentation on vKITTI · vKITTI 6-fold cross-validation | mAcc: 54.1 -> 67.3 | Helped |
| Replace unsupervised superpoint computation with SSP in SPG | SPG semantic segmentation on vKITTI · vKITTI 6-fold cross-validation | OA: 87.8 -> 84.3 | Hurt |
| Replace cross-partition weights with inversely proportional inter/intra-edge weights (ablation) | SSP · S3DIS, first cross-validation fold, 500 superpoints | OOA: -2.6 | Hurt |
| Replace cross-partition weights with inversely proportional inter/intra-edge weights (ablation) | SSP · S3DIS, first cross-validation fold, 500 superpoints | BR: -12.2 | Hurt |
| Replace cross-partition weights with inversely proportional inter/intra-edge weights (ablation) | SSP · S3DIS, first cross-validation fold, 500 superpoints | BP: +10.4 | Helped |
| Replace cross-partition weights with SEAL weights (ablation) | SSP · S3DIS, first cross-validation fold, 500 superpoints | OOA: -1.3 | Hurt |
| Replace cross-partition weights with SEAL weights (ablation) | SSP · S3DIS, first cross-validation fold, 500 superpoints | BR: -11.3 | Hurt |
| Replace cross-partition weights with SEAL weights (ablation) | SSP · S3DIS, first cross-validation fold, 500 superpoints | BP: +3.8 | Helped |
| Stack the local point embedder in two layers (ablation) | SSP · S3DIS, first cross-validation fold, 500 superpoints | OOA: -0.1 | No clear effect |
| Use two residual point embedder layers (ablation) | SSP · S3DIS, first cross-validation fold, 500 superpoints | OOA: +0.0 | No clear effect |
| Replace the loss functions with absolute distance and its negative (ablation) | SSP graph-structured contrastive loss · Point cloud oversegmentation | partition quality | Hurt |
Closest papers by meaning
- Large-Scale Point Cloud Semantic Segmentation With Superpoint Graphs (CVPR 2018 · similarity 0.90)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.87)
- Learning to Segment 3D Point Clouds in 2D Image Space (CVPR 2020 · similarity 0.87)
- Exploiting Local and Global Structure for Point Cloud Semantic Segmentation with Contextual Point Representations (NeurIPS 2019 · similarity 0.86)
- Mining Point Cloud Local Structures by Kernel Correlation and Graph Pooling (CVPR 2018 · similarity 0.86)
- Semantic Segmentation for Real Point Cloud Scenes via Bilateral Augmentation and Adaptive Fusion (CVPR 2021 · similarity 0.86)
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.