Research map / Point clouds / Segmentation and detection
Contrastive Boundary Learning for Point Cloud Segmentation
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 CBL to RandLA-Net | RandLA-Net · Semantic3D reduced-8 | mIoU: 77.4 -> 78.4 | Helped |
| Add CBL to ConvNet | ConvNet · S3DIS Area 5 | mIoU: 67.4 -> 69.4 | Helped |
| Add CBL to ConvNet | ConvNet · S3DIS 6-fold | mIoU: 69.7 -> 73.1 | Helped |
| Add CBL to ConvNet | ConvNet · ScanNet | mIoU: 69.1 -> 70.5 | Helped |
| Add CBL at input and sub-scenes rather than at input only (ablation) | ConvNet · ScanNet validation | mIoU: 70.05 -> 70.98 | Helped |
| Add a multiscale head without CBL (ablation) | ConvNet · ScanNet validation | OA: 88.97 -> 88.88 | Hurt |
| Add a multiscale head to ConvNet with CBL (ablation) | ConvNet + CBL · ScanNet validation | mIoU: 70.98 -> 71.33 | Helped |
| Set CBL temperature to 1 instead of 10 (ablation) | ConvNet + CBL · ScanNet validation | mIoU: 70.03 -> 71.33 | Helped |
| Use nearest-point labels for sub-sampled points instead of pooled label distributions (ablation) | ConvNet + CBL · S3DIS Area 5 | mIoU: 69.4 -> 68.3 | Hurt |
| Replace pooled label distributions with hard argmax labels (ablation) | ConvNet + CBL · S3DIS Area 5 | mIoU: 69.4 -> 66.8 | Hurt |
| Use KL-distance thresholding for sub-scene boundary determination (ablation) | ConvNet + CBL · S3DIS Area 5 | mIoU: 69.4 -> 69.5 | Helped |
| Add CBL to baselines | RandLA-Net, CloserLook3D, and ConvNet · S3DIS Area 5 boundary and inner areas | mIoU: boundary: +3.3, +0.6, +2.5; inner: +1.4, +0.2, +2.0 | Helped |
Closest papers by meaning
- Point Cloud Pre-training with Diffusion Models (CVPR 2024 · similarity 0.87)
- Hyperbolic Uncertainty-Aware Few-Shot Incremental Point Cloud Segmentation (CVPR 2025 · similarity 0.87)
- X-3D: Explicit 3D Structure Modeling for Point Cloud Recognition (CVPR 2024 · similarity 0.86)
- Multi-Modality Affinity Inference for Weakly Supervised 3D Semantic Segmentation (AAAI 2024 · similarity 0.86)
- Generalized Few-shot 3D Point Cloud Segmentation with Vision-Language Model (CVPR 2025 · similarity 0.86)
- FreePoint: Unsupervised Point Cloud Instance Segmentation (CVPR 2024 · 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.