AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Viewpoint Invariant Dense Matching for Visual Geolocalization

Gabriele Berton, Carlo Masone, Valerio Paolicelli, Barbara Caputo

ICCV 2021 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Invariance-based self-supervised learning · 40 citations (OpenAlex)

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 changedWhereMeasuredEffect
add GeoWarp re-ranking to GeMAlexNet GeM · R-Tokyo 25mrecall@1: 23.2 -> 34.0Improved
add GeoWarp re-ranking to NetVLADResNet-50 NetVLAD · R-Tokyo 25mrecall@1: 72.4 -> 74.4Improved
add GeoWarp re-ranking to NetVLADVGG16 NetVLAD · Pitts30k 25mrecall@1: 82.5 -> 83.3Improved
add self-supervised loss to warping training (ablation)ResNet50 NetVLAD GeoWarp · Pitts30k valrecall@1: 80.7 -> 87.0Improved
re-rank with local features instead of global features, untrained (ablation)ResNet50 NetVLAD · Pitts30k valrecall@1: 86.4 -> 80.7Hurt
add consistency loss alone, without self-supervision (ablation)VGG16 GeM GeoWarp · Pitts30k valrecall@1: 74.3 -> 74.2No clear effect
add feature-weighted weak loss alone, without self-supervision (ablation)VGG16 GeM GeoWarp · Pitts30k valrecall@1: 74.3 -> 74.3No clear effect
add both weak losses on top of self-supervised loss (ablation)VGG16 GeM GeoWarp · Pitts30k valrecall@1: 78.6 -> 79.0Improved
add QE + DBA query expansion to NetVLADResNet-50 NetVLAD · Pitts30k 25mrecall@1: 84.3 -> 83.7Hurt
add InLoc re-ranking instead of NetVLAD aloneVGG16 NetVLAD · Pitts30k 25mrecall@1: 82.5 -> 72.8Hurt
use Rocco et al. warping instead of GeoWarp warping moduleVGG16 NetVLAD · Pitts30k 25mrecall@1: 83.3 -> 81.9Hurt
increase warping coefficient k (ablation)GeoWarp · coarser thresholdsrecall@1Mixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy