Research map / 3D vision / 3D generation
Mapillary Street-Level Sequences: A Dataset for Lifelong Place Recognition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train NetVLAD on the diverse MSLS dataset instead of Pittsburgh250k | NetVLAD · MSLS test set, im2im | recall@5: 0.35 -> 0.58 | Helped |
| Train NetVLAD on MSLS rather than Pittsburgh250k | NetVLAD + CAT · MSLS test set, seq2seq | recall@5: 0.33 -> 0.60 | Helped |
| Train NetVLAD on MSLS rather than Pittsburgh250k | NetVLAD + MIN · MSLS test set, seq2im | recall@5: 0.37 -> 0.68 | Helped |
| Train NetVLAD on MSLS rather than Pittsburgh250k | NetVLAD + MIN · MSLS test set, im2seq | recall@5: 0.20 -> 0.34 | Helped |
| Concatenate rather than max-pool sequence embeddings (ablation) | NetVLAD · MSLS test set, seq2seq | recall@5: 0.58 -> 0.60 | Helped |
| Concatenate rather than average-pool sequence embeddings (ablation) | NetVLAD · MSLS test set, seq2seq | recall@5: 0.56 -> 0.60 | Helped |
| Use max pooling rather than average pooling across GeM sequence embeddings (ablation) | GeM · MSLS test set, seq2seq | recall@5: 0.42 -> 0.43 | No clear effect |
| Select the closest database image rather than the most frequent nearest image (ablation) | NetVLAD sequence-to-image recognition · MSLS test set, seq2im | recall@5: 0.37 -> 0.68 | Helped |
| Select the closest database image rather than the most frequent nearest image (ablation) | GeM sequence-to-image recognition · MSLS test set, seq2im | recall@5: 0.45 -> 0.47 | Helped |
| Use simple pooling strategies to model sequences (ablation) | MultiViewNet variants · MSLS test set, sequence-based recognition | recall | No clear effect |
Closest papers by meaning
- USCILab3D: A Large-scale, Long-term, Semantically Annotated Outdoor Dataset (NeurIPS 2024 · similarity 0.85)
- GeoFidelity-Bench: Evaluating Block-Conditioned Geographic Fidelity of Street-View Generation (NeurIPS 2026 · similarity 0.85)
- 3D StreetUnveiler with Semantic-aware 2DGS - a simple baseline (ICLR 2025 · similarity 0.85)
- OpenStreetView-5M: The Many Roads to Global Visual Geolocation (CVPR 2024 · similarity 0.85)
- L2RSI: Cross-view LiDAR-based Place Recognition for Large-scale Urban Scenes via Remote Sensing Imagery (NeurIPS 2025 · similarity 0.85)
- GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization (NeurIPS 2025 · 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.