Research map / Label-efficient and robust learning / Recognition and long-tail
Towards Test-time Efficient Visual Place Recognition via Asymmetric Query Processing
Jaeyoon Kim, Yoonki Cho, Sung-Eui Yoon
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 implicit embedding augmentation to asymmetric contrastive loss (ablation) | asymmetric VPR (BoQ gallery, EfficientViT-B2 query) · Nordland | Recall@1: 70.3 -> 74.6 | Improved |
| add implicit embedding augmentation (ablation) | asymmetric VPR (BoQ gallery, EfficientViT-B2 query) · Tokyo24/7 | Recall@1: 90.2 -> 92.7 | Improved |
| use implicit instead of explicit (K=10) embedding augmentation (ablation) | asymmetric VPR (BoQ gallery, EfficientViT-B2 query) · Nordland | Recall@1: 73.3 -> 74.6 | Improved |
| use geographical memory bank instead of queue-based memory bank (ablation) | asymmetric VPR (BoQ gallery, EfficientViT-B2 query) · Nordland | Recall@1: 72.0 -> 74.6 | Improved |
| use same aggregator type as gallery in query network instead of MLP transformation (ablation) | asymmetric VPR · Tokyo24/7 | Recall@1: 89.1 -> 92.7 | Improved |
| use geographical memory bank instead of k-NN (CSD) for compatible training | asymmetric VPR · Gallery embedding precomputation time | minutes: 1392.76 -> 0.26 | Improved |
| use geographical memory bank instead of k-NN (CSD) for compatible training | asymmetric VPR · training GPU memory | GB: 23.9 -> 17.5 | Improved |
| use asymmetric training with proposed method instead of symmetric lightweight model | EfficientViT-B2 query with BoQ gallery · Nordland | Recall@1: 51.6 -> 74.6 | Improved |
| use proposed method versus best k-NN baseline on Pitts250k with MobileViTv2/BoQ | MobileViTv2 query with BoQ gallery · Pitts250k | Recall@1: 93.2 -> 93.4 | No clear effect |
| replace DINOv2-B query with EfficientViT-B2 query (SALAD) | SALAD query network · query FLOPs | relative FLOPs: 100% -> 7.3% | Improved |
| use lightweight query network instead of high-capacity gallery model | BoQ asymmetric VPR · Pitts250k | Recall@1: 96.6 -> 95.4 | 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.
- D3still — D3still: Decoupled Differential Distillation for Asymmetric Image Retrieval (CVPR 2024) · better on 10 of 10 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
- EffoVPR: Effective Foundation Model Utilization for Visual Place Recognition (ICLR 2025 · similarity 0.91)
- Focus on Local: Finding Reliable Discriminative Regions for Visual Place Recognition (AAAI 2025 · similarity 0.90)
- Optimal Transport Aggregation for Visual Place Recognition (CVPR 2024 · similarity 0.90)
- Towards Implicit Aggregation: Robust Image Representation for Place Recognition in the Transformer Era (NeurIPS 2025 · similarity 0.89)
- MutualVPR: A Mutual Learning Framework for Resolving Supervision Inconsistencies via Adaptive Clustering (NeurIPS 2025 · similarity 0.88)
- Towards Seamless Adaptation of Pre-trained Models for Visual Place Recognition (ICLR 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.
