Research map / Label-efficient and robust learning / Recognition and long-tail
Multi-Label Learning with Contrastive Cluster Self-Supervision for 3D Hierarchical Semantic Segmentation
Shuyu Cao, Chongshou Li, Jie Xu, Tianrui Li, Na Zhao
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 |
|---|---|---|---|
| use late-decoupled decoders per hierarchy instead of a shared decoder (ablation) | ML3DHS · S3DIS-H, PointNet++ | Avg. mIoU (%): 62.17 -> 66.43 | Improved |
| add coarse-to-fine hierarchical guidance (ablation) | ML3DHS · Campus3D, PointNet++ | Avg. mIoU (%): 59.96 -> 63.28 | Improved |
| add contrastive discrimination branch (ablation) | ML3DHS · S3DIS-H, PointNet++ | Avg. mIoU (%): 62.98 -> 66.43 | Improved |
| add contrastive loss L_con (ablation) | ML3DHS · S3DIS-H, PointNet++ | Avg. mIoU (%): 61.49 -> 66.43 | Improved |
| add cross-hierarchy consistency loss L_chc (ablation) | ML3DHS · S3DIS-H, PointNet++ | Avg. mIoU (%): 64.09 -> 66.43 | Improved |
| add bi-branch self-supervised loss L_bis (ablation) | ML3DHS · S3DIS-H, PointNet++ | Avg. mIoU (%): 62.98 -> 66.43 | Improved |
| replace DHL with ML3DHS on PointNet++ | DHL · S3DIS-H | Avg. mIoU (%): 63.05 -> 66.43 | Improved |
| plug ML3DHS structure and losses into MTHS | MTHS · SensatUrban-H, PointNet++ | Avg. mIoU (%): 47.20 -> 49.05 | Improved |
| plug ML3DHS structure and losses into DHL | DHL · S3DIS-H, Point Transformer v3 | Avg. mIoU (%): 72.83 -> 75.34 | Improved |
| add auxiliary contrastive branch (training time) | ML3DHS · time per epoch, PointNet++ | training time | Hurt |
| vary auxiliary loss weight lambda in [0.1, 1.0] (ablation) | ML3DHS · hyper-parameter analysis | Avg. mIoU (%) | No clear effect |
Closest papers by meaning
- Hierarchical Intra-modal Correlation Learning for Label-free 3D Semantic Segmentation (CVPR 2024 · similarity 0.88)
- TWIST: Two-Way Inter-Label Self-Training for Semi-Supervised 3D Instance Segmentation (CVPR 2022 · similarity 0.87)
- GroupContrast: Semantic-aware Self-supervised Representation Learning for 3D Understanding (CVPR 2024 · similarity 0.85)
- AdaCo: Overcoming Visual Foundation Model Noise in 3D Semantic Segmentation via Adaptive Label Correction (AAAI 2025 · similarity 0.85)
- Flattening the Parent Bias: Hierarchical Semantic Segmentation in the Poincare Ball (CVPR 2024 · similarity 0.85)
- Multi-Label Prototype Visual Spatial Search for Weakly Supervised Semantic Segmentation (CVPR 2025 · 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.
