Research map / 3D vision / Depth and stereo
Sequential Visual Place Recognition: Exploiting Trajectory Priors for Robust Localization
Dominik Kloepfer, Patrick Wenzel
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 TRAIL CRF trajectory post-processing to MegaLoc, T=10 | MegaLoc · MSLS (Amsterdam+Boston) | Recall@T=10 (%): 58.3 -> 66.7 | Improved |
| add TRAIL, T=5 | MegaLoc · MSLS (Amsterdam+Boston) | Recall@T=5 (%): 58.0 -> 65.2 | Improved |
| add TRAIL, T=5 | MegaLoc · Nordland (winter->summer) | Recall@T=5 (%): 73.8 -> 81.3 | Improved |
| add TRAIL, T=10 | MegaLoc · 4Seasons countryside | Recall@T=10 (%): 68.6 -> 79.0 | Improved |
| add TRAIL with only the first image (T=1) | MegaLoc · MSLS (Amsterdam+Boston) | Recall@T=1 (%): 55.1 -> 56.5 | No clear effect |
| add learned transition modeling instead of Location Filtering heuristic | Location Filtering · MSLS (Amsterdam+Boston) | Recall@T=10 (%): 58.3 -> 66.7 | Improved |
| add sequence-descriptor retrieval (CaseVPR) to unordered database | MegaLoc · MSLS (Amsterdam+Boston) | Recall@T=1 (%): 55.1 -> 33.3 | Hurt |
| add Majority Voting heuristic, T=5 | MegaLoc · Nordland (winter->summer) | Recall@T=5 (%): 73.8 -> 19.5 | Hurt |
| add Max. Similarity heuristic, T=5 | MegaLoc · MSLS (Amsterdam+Boston) | Recall@T=5 (%): 58.0 -> 39.1 | Hurt |
| add Location Filtering heuristic, T=5 | MegaLoc · Nordland (winter->summer) | Recall@T=5 (%): 73.8 -> 79.7 | Improved |
| replace scaled sigmoid kappa with supergaussian (ablation) | TRAIL · MSLS | Recall | No clear effect |
| use larger ViT-L backbone (FoL) instead of TRAIL on ViT-B | TRAIL · Nordland (winter->summer) | Recall@T=10 (%): 78.1 -> 55.5 | Hurt |
Closest papers by meaning
- Focus on Local: Finding Reliable Discriminative Regions for Visual Place Recognition (AAAI 2025 · similarity 0.89)
- RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations (CVPR 2025 · similarity 0.88)
- Towards Implicit Aggregation: Robust Image Representation for Place Recognition in the Transformer Era (NeurIPS 2025 · similarity 0.88)
- Optimal Transport Aggregation for Visual Place Recognition (CVPR 2024 · similarity 0.88)
- MutualVPR: A Mutual Learning Framework for Resolving Supervision Inconsistencies via Adaptive Clustering (NeurIPS 2025 · similarity 0.88)
- On the Estimation of Image-matching Uncertainty in Visual Place Recognition (CVPR 2024 · similarity 0.88)
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.
