$R^{2}$Former: Unified $R$etrieval and $R$eranking Transformer for Place Recognition

Benchmark Model Rank Results
visual-place-recognition-on-st-luciaR2Former (ViT-Small, MSLS-trained, top-100 reranking)#7Recall@1: 99.7
visual-place-recognition-on-mapillary-testR2Former (ViT-Small, MSLS-trained, top-100 reranking)#16Recall@1: 73Recall@5: 85.9Recall@10: 88.8
visual-place-recognition-on-tokyo247R2Former (ViT-Small, Pitts30k-finetuned, top-100 reranking)#19Recall@1: 88.6
visual-place-recognition-on-mapillary-valR2Former (ViT-Base, MSLS-trained, top-100 reranking)#23Recall@1: 90Recall@5: 95.1Recall@10: 96.9
visual-place-recognition-on-pittsburgh-30k-testR2Former (ViT-Small, Pitts30k-finetuned, top-100 reranking)#24Recall@1: 91.1Recall@5: 95.2Recall@10: 96.3