Research map / 3D vision / Depth and stereo
Unordered Landmark Visual Navigation
Hao Ren, Junzhe Zhu, Yihan Li, Zetong Bi, Le Zheng, Zhi Li, Yiqing Yuan, Zhaoliang Wan, Dizhe Zhang, Lu Qi, HUI CHENG
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 maximum spanning forest refinement (ablation) | RAVEL · GRScenes mapping | F1: 0.5157 -> 0.7365 | Improved |
| add adaptive verification threshold (AVP tau) (ablation) | RAVEL w/o MSF · GRScenes mapping | F1: 0.4496 -> 0.5157 | Improved |
| add geometric verification and MSF on top of top-k retrieval (ablation) | Top-k ANN · GRScenes mapping | F1: 0.2156 -> 0.7365 | Improved |
| raise retrieval top-k from 2 to 5 (ablation) | PlaceNav · GRScenes mapping | F1: 0.6043 -> 0.3731 | Hurt |
| add entropy-adaptive fusion (ablation) | BPL · GRScenes localization | Accuracy (%): 90.57 -> 95.49 | Improved |
| increase propagation depth k from 1 to 5 (ablation) | BPL · GRScenes localization | Accuracy (%): 95.49 -> 92.73 | Hurt |
| replace temporal-distance localization with BPL (ablation) | ULVN+ViNT · GRScenes navigation | Success rate (%): 59.6 -> 68.1 | Improved |
| add temporal distance loss in local planner training (ablation) | ULVN+ViNT · GRScenes navigation | Success rate (%): 31.0 -> 68.1 | Improved |
| add temporal distance loss in local planner training (ablation) | ULVN+NoMaD · GRScenes navigation | Success rate (%): 54.3 -> 71.9 | Improved |
| use BPL graph-based belief propagation instead of ViNT temporal distance | localization · GRScenes difficult paths | Accuracy (%): 70.50 -> 93.99 | Improved |
| add image noise perturbation | RAVEL · GRScenes mapping | F1: 0.7365 -> 0.6882 | No clear effect |
| add image noise perturbation | ViNT top 5 · GRScenes mapping | F1: 0.3806 -> 0.3281 | Hurt |
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.
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · 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
- Change-Robust Online Topological Memory for Long-Term Relocalization and Semantic Navigation (NeurIPS 2026 · similarity 0.86)
- UniGoal: Towards Universal Zero-shot Goal-oriented Navigation (CVPR 2025 · similarity 0.86)
- BeliefMapNav: 3D Voxel-Based Belief Map for Zero-Shot Object Navigation (NeurIPS 2025 · similarity 0.85)
- Seeing through Uncertainty: Robust Task-Oriented Optimization in Visual Navigation (NeurIPS 2025 · similarity 0.85)
- MemoNav: Working Memory Model for Visual Navigation (CVPR 2024 · similarity 0.84)
- Instance-aware Exploration-Verification-Exploitation for Instance ImageGoal Navigation (CVPR 2024 · 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.
