Research map / Label-efficient and robust learning / Test-time adaptation
MM-TTA: Multi-Modal Test-Time Adaptation for 3D Semantic Segmentation
Inkyu Shin, Yi-Hsuan Tsai, Bingbing Zhuang, Samuel Schulter, Buyu Liu, Sparsh Garg, In So Kweon, Kuk-Jin Yoon
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 |
|---|---|---|---|
| use MM-TTA soft selection instead of the strongest reported baseline | 3D semantic segmentation · A2D2 → SemanticKITTI | mIoU (Softmax avg): 49.1 -> 47.1 | Hurt |
| use MM-TTA soft selection instead of the strongest reported baseline | 3D semantic segmentation · Synthia → SemanticKITTI | mIoU (Softmax avg): 35.3 -> 35.1 | Hurt |
| use MM-TTA soft selection instead of the strongest reported baseline | 3D semantic segmentation · nuScenes Day → Night | mIoU (Softmax avg): 50.8 -> 51.8 | Improved |
| add the slowly updated model to fast-model pseudo-label generation (ablation) | MM-TTA Intra-PG · A2D2 → SemanticKITTI | mIoU (Softmax avg): 40.8 -> 41.9 | Improved |
| add pseudo-label thresholding to consensus fusion (ablation) | MM-TTA Inter-PR · A2D2 → SemanticKITTI | mIoU (Softmax avg): 41.8 -> 43.8 | Improved |
| use mean-probability merging with pseudo-label thresholding (ablation) | MM-TTA · A2D2 → SemanticKITTI | mIoU (Softmax avg): 41.6 -> 45.3 | Improved |
| add the slowly updated model to merge and thresholding (ablation) | MM-TTA · A2D2 → SemanticKITTI | mIoU (Softmax avg): 41.6 -> 45.3 | Improved |
| select pseudo-labels using prediction entropy rather than the proposed consistency measure (ablation) | MM-TTA Inter-PR · A2D2 → SemanticKITTI | mIoU (Softmax avg): 43.4 -> 47.1 | Hurt |
| use consistency-based hard selection instead of consistency-weighted soft selection (ablation) | MM-TTA Inter-PR · A2D2 → SemanticKITTI | mIoU (Softmax avg): 47.0 -> 47.1 | No clear effect |
| set the hard-selection pseudo-label threshold to 0.1 rather than 0.3 (ablation) | MM-TTA (Hard Select) · A2D2 → SemanticKITTI | mIoU (Softmax avg): 47.0 -> 45.2 | Hurt |
| set the hard-selection pseudo-label threshold to 0.5 rather than 0.3 (ablation) | MM-TTA (Hard Select) · A2D2 → SemanticKITTI | mIoU (Softmax avg): 47.0 -> 46.7 | Hurt |
| set the hard-selection pseudo-label threshold to 0.7 rather than 0.3 (ablation) | MM-TTA (Hard Select) · A2D2 → SemanticKITTI | mIoU (Softmax avg): 47.0 -> 46.2 | Hurt |
Closest papers by meaning
- Multi-Modal Continual Test-Time Adaptation for 3D Semantic Segmentation (ICCV 2023 · similarity 0.91)
- Cross-Modal Learning for Domain Adaptation in 3D Semantic Segmentation (TPAMI 2022 · similarity 0.89)
- Unsupervised Modality Adaptation with Text-to-Image Diffusion Models for Semantic Segmentation (NeurIPS 2024 · similarity 0.89)
- Test-Time Adaptation of Vision-Language Models for Open-Vocabulary Semantic Segmentation (NeurIPS 2025 · similarity 0.87)
- Hierarchical Intra-modal Correlation Learning for Label-free 3D Semantic Segmentation (CVPR 2024 · similarity 0.86)
- AdaCo: Overcoming Visual Foundation Model Noise in 3D Semantic Segmentation via Adaptive Label Correction (AAAI 2025 · 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.
