Research map / Point clouds / Segmentation and detection
CGA-Net: Category Guided Aggregation for Point Cloud Semantic Segmentation
Tao Lu, Limin Wang, Gangshan Wu
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 CGA feature augmentation | CloserLook3D(pseudo-grid) · S3DIS Area 5 | mIoU(%): 65.7 -> 68.6 | Improved |
| Add CGA feature augmentation | RandLA-Net · S3DIS Area 5 | mIoU(%): 62.5 -> 65.4 | Improved |
| Add CGA feature augmentation | MeteorNet · Synthia | mIoU(%): 81.80 -> 83.75 | Improved |
| Add same-category aggregation (SAM) (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 64.9 -> 68.6 | Improved |
| Add different-category relation module (DRM) (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 66.5 -> 68.6 | Improved |
| Add the supervised binary classification loss (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 66.7 -> 68.6 | Improved |
| Use supervised binary classification instead of cosine similarity (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 66.6% -> 68.6 | Improved |
| Use predicted binary labels instead of ground-truth binary labels (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 70.1% -> 68.6 | Hurt |
| Replace SAM with DRM (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 68.6 -> 67.2 | Hurt |
| Replace DRM with SAM (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 68.6 -> 66.2 | Hurt |
| Use subtraction instead of concatenation in DRM (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 67.5 -> 68.6 | Improved |
| Use subtraction instead of addition in DRM (ablation) | CGA with CloserLook3D · S3DIS Area 5 | mIoU(%): 67.4 -> 68.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- RandLA-Net — RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds (CVPR 2020) · better on 1 of 1 shared result
- JSENet — JSENet: Joint Semantic Segmentation and Edge Detection Network for 3D Point Clouds (ECCV 2020) · better on 1 of 1 shared result
- MeteorNet — MeteorNet: Deep Learning on Dynamic 3D Point Cloud Sequences (ICCV 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Subspace Prototype Guidance for Mitigating Class Imbalance in Point Cloud Semantic Segmentation (ECCV 2024 · similarity 0.87)
- SUM Parts: Benchmarking Part-Level Semantic Segmentation of Urban Meshes (CVPR 2025 · similarity 0.84)
- Rethinking Few-shot 3D Point Cloud Semantic Segmentation (CVPR 2024 · similarity 0.84)
- Distribution Guidance Network for Weakly Supervised Point Cloud Semantic Segmentation (NeurIPS 2024 · similarity 0.84)
- Semantic-aware SAM for Point-Prompted Instance Segmentation (CVPR 2024 · similarity 0.84)
- Weakly Supervised Point Cloud Semantic Segmentation via Artificial Oracle (CVPR 2024 · 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.
