Research map / Point clouds / Point cloud classification
Partially Matching Submap Helps: Uncertainty Modeling and Propagation for Text to Point Cloud Localization
Mingtao Feng, Longlong Mei, Zijie Wu, Jianqiao Luo, Fenghao Tian, Jie Feng, Weisheng Dong, Yaonan Wang
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 uncertainty-based NLL loss when training with partially matching submaps (vs plain regression) (ablation) | fine localization · KITTI360Pose test k=1 eps<5m | localization recall: 0.21 -> 0.39 | Improved |
| add uncertainty-based NLL loss when training with partially matching submaps (vs plain regression) (ablation) | fine localization · KITTI360Pose val k=1 eps<5m | localization recall: 0.25 -> 0.42 | Improved |
| add partially matching submaps to Text2Loc contrastive training naively (ablation) | coarse place recognition · KITTI360Pose test k=1 | submap retrieval recall: 0.28 -> 0.16 | Hurt |
| add L2-distance assignment loss for partially matching submaps (ablation) | coarse place recognition · KITTI360Pose test k=1 | submap retrieval recall: 0.16 -> 0.31 | Improved |
| use IOU-based assignment loss instead of L2-distance loss (ablation) | coarse place recognition · KITTI360Pose test k=1 | submap retrieval recall: 0.31 -> 0.32 | Improved |
| propagate uncertainty into IOU assignment loss (L'iou) (ablation) | coarse place recognition · KITTI360Pose test k=1 | submap retrieval recall: 0.32 -> 0.34 | Improved |
| use our coarse retrieval for Text2Loc fine localization (ablation) | Text2Loc · KITTI360Pose test k=1 eps<5m | localization recall: 0.33 -> 0.36 | Improved |
| add our partial-sample uncertainty framework to fine localization (ablation) | Text2Loc* · KITTI360Pose test k=1 eps<5m | localization recall: 0.36 -> 0.39 | Improved |
| train with partially and fully matching samples with uncertainty modeling | Text2Loc · KITTI360Pose | normalized Euclidean error (test): 0.090 -> 0.081 | Improved |
| train with partially and fully matching samples with uncertainty modeling | Text2Loc · CityRefer val k=1 eps<10m | localization recall: 0.25 -> 0.29 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Text2Loc — Text2Loc: 3D Point Cloud Localization from Natural Language (CVPR 2024) · better on 9 of 9 shared results
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
- Text2Loc: 3D Point Cloud Localization from Natural Language (CVPR 2024 · similarity 0.89)
- Text to Point Cloud Localization with Multi-Level Negative Contrastive Learning (AAAI 2025 · similarity 0.89)
- CMMLoc: Advancing Text-to-PointCloud Localization with Cauchy-Mixture-Model Based Framework (CVPR 2025 · similarity 0.88)
- Semantic Ambiguity Modeling and Propagation for Fine-Grained Visual Cross View Geo-Localization (AAAI 2025 · similarity 0.87)
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.87)
- On the Estimation of Image-matching Uncertainty in Visual Place Recognition (CVPR 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.
