ProGEO: Generating Prompts through Image-Text Contrastive Learning for Visual Geo-localization

Benchmark Model Rank Results
visual-place-recognition-on-mslsProGEO#1Recall@1: 84.9Recall@5: 91.6
visual-place-recognition-on-pittsburgh-250kProGEO#10Recall@1: 92.2Recall@5: 97.7
visual-place-recognition-on-pittsburgh-30kProGEO#7Recall@1: 93.0Recall@5: 98.3
visual-place-recognition-on-sf-xl-test-v1ProGEO#1Recall@1: 84.7Recall@5: 90.3
visual-place-recognition-on-sf-xl-test-v2ProGEO#1Recall@1: 93.0Recall@5: 96.7
visual-place-recognition-on-st-luciaProGEO#5Recall@1: 99.7Recall@5: 99.9
visual-place-recognition-on-tokyo247ProGEO#8Recall@1: 88.6Recall@5: 93.3