Research map / Point clouds / Segmentation and detection
LASS3D: Language-Assisted Semi-Supervised 3D Semantic Segmentation with Progressive Unreliable Data Exploitation
Jianan Li, Qiulei Dong
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 unlabeled CE loss from reliable pseudo labels (ablation) | LASS3D · SemanticKITTI 1% | mIoU: 49.6 -> 53.7 | Improved |
| add KL distillation of text-enhanced features to teacher (ablation) | LASS3D · SemanticKITTI 1% | mIoU: 53.7 -> 55.6 | Improved |
| add progressive negative learning on unreliable points (ablation) | LASS3D · SemanticKITTI 1% | mIoU: 55.6 -> 58.5 | Improved |
| add both image-level and entity-level captions (ablation) | LASS3D · SemanticKITTI 1% | mIoU: 50.0 -> 58.5 | Improved |
| use cross attention instead of add or concatenation to fuse image-level embeddings (ablation) | LASS3D · SemanticKITTI 1% | mIoU: 56.2 -> 58.5 | Improved |
| use SAFM fusion instead of alignment-based CLIP2Scene regularization (ablation) | LASS3D · SemanticKITTI 1% | mIoU: 53.0 -> 58.5 | Improved |
| use dual-score (confidence plus variance) separation instead of confidence only (ablation) | LASS3D · SemanticKITTI 1% | mIoU: 57.7 -> 58.5 | Improved |
| integrate progressive exploitation into GPC (ablation) | GPC · SemanticKITTI 1% | mIoU: 54.1 -> 54.7 | Improved |
| increase negative learning iterations beyond about 5 (ablation) | LASS3D · SemanticKITTI | mIoU | No clear effect |
| LASS3D versus IGNet at 50% labeled | LASS3D · SemanticKITTI 50% | mIoU: 64.8 -> 64.5 | Hurt |
Closest papers by meaning
- Bayesian Self-Training for Semi-Supervised 3D Segmentation (ECCV 2024 · similarity 0.90)
- Density-Guided Semi-Supervised 3D Semantic Segmentation with Dual-Space Hardness Sampling (CVPR 2024 · similarity 0.89)
- BFANet: Revisiting 3D Semantic Segmentation with Boundary Feature Analysis (CVPR 2025 · similarity 0.88)
- 3D-AVS: LiDAR-based 3D Auto-Vocabulary Segmentation (CVPR 2025 · similarity 0.88)
- Segment Any 3D Object with Language (ICLR 2025 · similarity 0.87)
- Multi-Modality Affinity Inference for Weakly Supervised 3D Semantic Segmentation (AAAI 2024 · similarity 0.87)
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.
