Research map / Point clouds / Segmentation and detection
Investigate Indistinguishable Points in Semantic Segmentation of 3D Point Cloud
Mingye Xu, Zhipeng Zhou, Junhao Zhang, Yu Qiao
Read the paperCheck 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 IAF module (ablation) | IAF-Net baseline · S3DIS Area 5 | mIoU (%): 62.60 -> 64.60 | Improved |
| Add IAF module (ablation) | IAF-Net · S3DIS Area 5 | mIoU (%): 62.6 -> 64.6 | Improved |
| Replace attentive pooling with max-pooling while removing IAF (ablation) | IAF-Net · S3DIS Area 5 | mIoU (%): 62.6 -> 60.6 | Hurt |
| Remove correlations from the no-IAF, max-pooling variant (ablation) | IAF-Net · S3DIS Area 5 | mIoU (%): 60.6 -> 59.9 | Hurt |
| Remove the encoder multi-scale strategy (ablation) | IAF-Net · S3DIS Area 5 | mIoU (%): 64.6 -> 59.2 | Hurt |
| Use fine-grained-feature local difference alone for mining (ablation) | IAF-Net · S3DIS Area 5 | m-IoU (%): 61.7 -> 64.6 | Improved |
| Use original-point local difference alone instead of fine-grained-feature difference (ablation) | IAF-Net · S3DIS Area 5 | m-IoU (%): 64.6 -> 61.7 | Hurt |
| Use semantic-prediction local difference alone instead of fine-grained-feature difference (ablation) | IAF-Net · S3DIS Area 5 | m-IoU (%): 64.6 -> 62.1 | Hurt |
| Combine all three local differences equally instead of using fine-grained-feature difference alone (ablation) | IAF-Net · S3DIS Area 5 | m-IoU (%): 64.6 -> 63.7 | Hurt |
| Add multi-stage supervision (ablation) | IAF-Net · S3DIS Area 5 | m-IoU (%): 57.7 -> 64.6 | Improved |
| Add stage-5 supervision to the multi-stage loss (ablation) | IAF-Net · S3DIS Area 5 | m-IoU (%): 61.9 -> 64.6 | Improved |
| Use IAF-Net instead of KPConv | S3DIS Area 5 · S3DIS Area 5 | mIoU (%): 67.1 -> 64.6 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- Point2Node — Point2Node: Correlation Learning of Dynamic-Node for Point Cloud Feature Modeling (AAAI 2020) · better on 2 of 2 shared results
- RandLANet — RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds (CVPR 2020) · better on 1 of 1 shared result
- PointWeb — PointWeb: Enhancing Local Neighborhood Features for Point Cloud Processing (CVPR 2019) · better on 0 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
- Geometry and Uncertainty-Aware 3D Point Cloud Class-Incremental Semantic Segmentation (CVPR 2023 · similarity 0.89)
- PointDC: Unsupervised Semantic Segmentation of 3D Point Clouds via Cross-Modal Distillation and Super-Voxel Clustering (ICCV 2023 · similarity 0.87)
- Weakly Supervised Point Cloud Semantic Segmentation via Artificial Oracle (CVPR 2024 · similarity 0.86)
- Rethinking Few-shot 3D Point Cloud Semantic Segmentation (CVPR 2024 · similarity 0.86)
- Less Is More: Label Recommendation for Weakly Supervised Point Cloud Semantic Segmentation (AAAI 2024 · similarity 0.86)
- Multi-Modality Affinity Inference for Weakly Supervised 3D Semantic Segmentation (AAAI 2024 · similarity 0.85)
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.
