The authors propose distance-aware cross-view geo-localization, release the DA-Campus benchmark, and show a multi-scale contrastive loss with re-ranking improves both hierarchical and standard retrieval.
Building information modeling and classification by visual learning at a city scale,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Dynamic Contrastive Learning for Hierarchical Retrieval: A Case Study of Distance-Aware Cross-View Geo-Localization
The authors propose distance-aware cross-view geo-localization, release the DA-Campus benchmark, and show a multi-scale contrastive loss with re-ranking improves both hierarchical and standard retrieval.