Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• CORE is the first million-scale dataset dedicated to global cross-modal geo-localization.technical 95%
• CORE comprises 1,034,786 cross-view images sampled from 225 distinct geographic regions across all continents.technical 95%
• Existing cross-modal geo-localization research is constrained by narrow geographic coverage and simplistic scene diversity.limitation 85%
• The PLANET network significantly outperforms state-of-the-art methods in cross-modal geo-localization.result 90%
• Large Vision-Language Models are used to synthesize high-quality scene descriptions rich in discriminative cues for the CORE dataset.method 90%
• PLANET introduces a novel contrastive learning paradigm to guide textual representations in capturing physical signatures of satellite imagery.method 90%
• Cross-modal geo-localization is crucial for pedestrian navigation and emergency response applications.market 85%
• The CORE dataset offers unprecedented variety of perspectives in varying environmental conditions and urban layouts.technical 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card