Towards Implicit Aggregation: Robust Image Representation for Place Recognition in the Transformer Era

Benchmark Model Rank Results
visual-place-recognition-on-nordlandImAge (6144D, unified dataset)#1Recall@1: 97.7Recall@5: 99.3Recall@10: 99.6
visual-place-recognition-on-mapillary-testImAge#2Recall@1: 84.5Recall@5: 93.8Recall@10: 95.4
visual-place-recognition-on-svox-nightImAge (6144D, GSV-Cities)#4Recall@1: 97.6Recall@5: 99.4
visual-place-recognition-on-mapillary-valImAge (6144D, unified dataset)#5Recall@1: 94.5Recall@5: 97.3Recall@10: 98
visual-place-recognition-on-st-luciaImAge (6144D, GSV-Cities)#5Recall@1: 99.9Recall@5: 100
visual-place-recognition-on-svox-sunImAge (6144D, GSV-Cities)#5Recall@1: 98Recall@5: 99.5
visual-place-recognition-on-pittsburgh-250k-testImAge (6144D, GSV-Cities)#9Recall@1: 96.5Recall@5: 99.1
visual-place-recognition-on-pittsburgh-30k-testImAge (6144D, unified dataset)#10Recall@1: 94.1Recall@5: 97.3Recall@10: 98.1
visual-place-recognition-on-spedImAge (6144D, GSV-Cities)#10Recall@1: 91.6Recall@5: 95.6
visual-place-recognition-on-tokyo247ImAge (6144D, unified dataset)#13Recall@1: 97.1