Research map / Label-efficient and robust learning / Domain adaptation and generalization
xMUDA: Cross-Modal Unsupervised Domain Adaptation for 3D Semantic Segmentation
Maximilian Jaritz, Tuan-Hung Vu, Raoul de Charette, Emilie Wirbel, Patrick Perez
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 cross-modal learning to source-only training | 2D stream, USA/Singapore · USA/Singapore target test | mIoU | Improved |
| Add cross-modal learning to source-only training | 3D stream, USA/Singapore · USA/Singapore target test | mIoU | Improved |
| Add pseudo-label self-training to xMUDA | xMUDA · USA/Singapore target test | mIoU (softmax avg): 62.7 -> 63.2 | Improved |
| Add pseudo-label self-training to xMUDA | xMUDA · Day/Night target test | mIoU (softmax avg): 50.0 -> 50.8 | Improved |
| Add pseudo-label self-training to xMUDA | xMUDA · A2D2/SemanticKITTI target test | mIoU (softmax avg): 42.9 -> 49.1 | Improved |
| Add pseudo-label self-training to xMUDA | xMUDA · Day/Night target test | mIoU (softmax avg): 50.0 -> 50.8; xMUDA_PL does not achieve the best score | Mixed |
| Train the mimicry head with cross-modal loss on source as well as target, rather than target only (ablation) | xMUDA · USA/Singapore | mIoU (2D): 4.8 mIoU | Improved |
| Train the mimicry head with cross-modal loss on source as well as target, rather than target only (ablation) | xMUDA · USA/Singapore | mIoU (3D): 4.4 mIoU | Improved |
| Use separate main and mimicry segmentation heads instead of a single shared head (ablation) | xMUDA · USA/Singapore | mIoU | Improved |
| Add cross-modal loss during supervised oracle training (ablation) | Oracle 2D stream · USA/Singapore oracle | mIoU: 65.8 -> 66.4 | Improved |
| Add cross-modal loss during supervised oracle training (ablation) | Oracle 3D stream · USA/Singapore oracle | mIoU: 63.2 -> 63.8 | Improved |
| Add cross-modal loss during supervised oracle training (ablation) | Oracle softmax average · USA/Singapore oracle | mIoU: 71.1 -> 71.6 | Improved |
Later papers that compared with this method
4 later accepted papers report xMUDA as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- UniDxMD: Towards Unified Representation for Cross-Modal Unsupervised Domain Adaptation in 3D Semantic Segmentation (ICCV 2025 · better on 4 of 4 shared results)
- Mx2M: Masked Cross-Modality Modeling in Domain Adaptation for 3D Semantic Segmentation (AAAI 2023 · better on 3 of 3 shared results)
- BEV-DG: Cross-Modal Learning under Bird's-Eye View for Domain Generalization of 3D Semantic Segmentation (ICCV 2023 · better on 1 of 1 shared result)
- mDALU: Multi-Source Domain Adaptation and Label Unification With Partial Datasets (ICCV 2021 · 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
- UniDxMD: Towards Unified Representation for Cross-Modal Unsupervised Domain Adaptation in 3D Semantic Segmentation (ICCV 2025 · similarity 0.93)
- Cross-Modal Learning for Domain Adaptation in 3D Semantic Segmentation (TPAMI 2022 · similarity 0.92)
- Cross-Modal Contrastive Learning for Domain Adaptation in 3D Semantic Segmentation (AAAI 2023 · similarity 0.92)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.90)
- DADA: Depth-Aware Domain Adaptation in Semantic Segmentation (ICCV 2019 · similarity 0.90)
- UniDSeg: Unified Cross-Domain 3D Semantic Segmentation via Visual Foundation Models Prior (NeurIPS 2024 · similarity 0.89)
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.
