Image Retrieval on Revisited Paris (RPar) (Medium)
94.2mAPSuperGlobal
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SuperGlobalBackbone=ResNet-101, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=16002023.08 | 94.2 | — | |
| SuperGlobalBackbone=ResNet-101, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=8002023.08 | 94.1 | — | |
| SuperGlobalBackbone=ResNet-101, Retrieval Category=SuperGlobal retrieval and reranking, Reranking Setting=Rerank top 4002023.08 | 93.3 | — | |
| SuperGlobalBackbone=ResNet-101, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=4002023.08 | 93.3 | — | |
| SuperGlobalBackbone=ResNet-50, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=16002023.08 | 93.3 | — | |
| SuperGlobalBackbone=ResNet-50, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=8002023.08 | 93 | — | |
| SuperGlobalBackbone=ResNet-101, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 92.1 | — | |
| SuperGlobalBackbone=ResNet-101, Strategy=Global feature retrieval2023.08 | 92.1 | — | |
| SuperGlobalBackbone=ResNet-50, Retrieval Category=SuperGlobal retrieval and reranking, Reranking Setting=Rerank top 4002023.08 | 92 | — | |
| SuperGlobalBackbone=ResNet-50, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=4002023.08 | 92 | — | |
| ViTGaLBackbone=XCiT-S24, Training Dataset=GLDv2, Result Source=Original paper [25]2023.09 | 91.4 | — | |
| ViTGaLBackbone=XCiT-S24, Training Dataset=GLDv2, Result Source=Reproduced code2023.09 | 91.4 | — | |
| CVNetBackbone=ResNet-101, Retrieval Category=Global feature retrieval + Local feature reranking, Reranking Setting=Rerank top 4002023.08 | 91.2 | — | |
| CVNetBackbone=ResNet-101, Strategy=Global feature retrieval + Local feature reranking, Reranking Depth=4002023.08 | 91.2 | — | |
| DOLGBackbone=ResNet-101, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 91 | — | |
| DOLGBackbone=ResNet-101, Strategy=Global feature retrieval2023.08 | 91 | — | |
| DOLGBackbone=ResNet-101, Training Dataset=GLDv2, Result Source=Original paper [46]2023.09 | 91 | — | |
| DOLGBackbone=ResNet-101, Training Dataset=GLDv2, Result Source=Reproduced code2023.09 | 91 | — | |
| SuperGlobalBackbone=ResNet-50, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 90.5 | — | |
| CVNetBackbone=ResNet-50, Retrieval Category=Global feature retrieval + Local feature reranking, Reranking Setting=Rerank top 4002023.08 | 90.5 | — | |
| SuperGlobalBackbone=ResNet-50, Strategy=Global feature retrieval2023.08 | 90.5 | — | |
| CVNetBackbone=ResNet-50, Strategy=Global feature retrieval + Local feature reranking, Reranking Depth=4002023.08 | 90.5 | — | |
| CVNetBackbone=ResNet-101, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 90.3 | — | |
| CVNetBackbone=ResNet-101, Strategy=Global feature retrieval2023.08 | 90.3 | — | |
| DOLGBackbone=ResNet-50, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 89.8 | — | |
| DOLGBackbone=ResNet-50, Strategy=Global feature retrieval2023.08 | 89.8 | — | |
| RRT# Reranked images=400, Desc. version=R50-v2-clean2021.03 | 89.1 | — | |
| CVNetBackbone=ResNet-50, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 88.8 | — | |
| CVNetBackbone=ResNet-50, Strategy=Global feature retrieval2023.08 | 88.8 | — | |
| alphaQE + RRT# Reranked images=200, Desc. version=R50-v2-clean2021.03 | 88.5 | — | |
| RRT# Reranked images=200, Desc. version=R50-v2-clean2021.03 | 87.8 | — | |
| RRT# local desc.=500, # Reranked images=200, Desc. version=R50-v2-clean2021.03 | 87.8 | — | |
| DELGBackbone=ResNet-101, Retrieval Category=Global feature retrieval + Local feature reranking, Reranking Setting=GV rerank top 1002023.08 | 87.2 | — | |
| DELGBackbone=ResNet-101, Strategy=Global feature retrieval + Local feature reranking, Reranking Depth=1002023.08 | 87.2 | — | |
| RRTDesc. version=R50-v2-clean, # local desc.=500, Desc. dim.=1282021.03 | 86.7 | — | |
| alphaQEDesc. version=R50-v2-clean2021.03 | 86.7 | — | |
| RRT# Reranked images=100, Desc. version=R50-v2-clean2021.03 | 86.7 | — | |
| RRT# local desc.=500, # Reranked images=100, Desc. version=R50-v2-clean2021.03 | 86.7 | — | |
| DELGBackbone=ResNet-101, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 86.6 | — | |
| DELGBackbone=ResNet-101, Strategy=Global feature retrieval2023.08 | 86.6 | — | |
| RRT# Reranked images=400, Desc. version=R50-v12021.03 | 86.3 | — | |
| SuperGlueDesc. version=SuperPoint [17], # local desc.=1024, Desc. dim.=256, Global Retrieval Baseline=DELG R50-v2-clean2021.03 | 86.2 | — | |
| SuperGlueDesc. version=SuperPoint [17], # local desc.=500, Desc. dim.=256, Global Retrieval Baseline=DELG R50-v2-clean2021.03 | 85.9 | — | |
| DELG globalDesc. version=R50-v2-clean, # local desc.=02021.03 | 85.7 | — | |
| DELG globalDesc. version=R50-v2-clean2021.03 | 85.7 | — | |
| DELG global# local desc.=0, # Reranked images=0, Desc. version=R50-v2-clean2021.03 | 85.7 | — | |
| GV# local desc.=1000, # Reranked images=100, Desc. version=R50-v2-clean2021.03 | 85.7 | — | |
| DELGBackbone=ResNet-50, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 85.7 | — | |
| DELGBackbone=ResNet-50, Retrieval Category=Global feature retrieval + Local feature reranking, Reranking Setting=GV rerank top 1002023.08 | 85.7 | — | |
| DELGBackbone=ResNet-50, Strategy=Global feature retrieval2023.08 | 85.7 | — | |
| DELGBackbone=ResNet-50, Strategy=Global feature retrieval + Local feature reranking, Reranking Depth=1002023.08 | 85.7 | — | |
| alphaQE + RRT# Reranked images=200, Desc. version=R50-v12021.03 | 85.6 | — | |
| GV*# local desc.=1000, # Reranked images=200, Desc. version=R50-v2-clean2021.03 | 85.5 | — | |
| CiDeR-FTTrain Set=SfM-120k, Backbone=ResNet-101, Pooling=GeM, Loss=ArcFace, Fine-tuned=true, Dimension=20482024.04 | 84.5 | — | |
| RRT# Reranked images=200, Desc. version=R50-v12021.03 | 84.4 | — | |
| RRT# local desc.=500, # Reranked images=200, Desc. version=R50-v12021.03 | 84.4 | — | |
| alphaQEDesc. version=R50-v12021.03 | 83.4 | — | |
| RRTDesc. version=R50-v1, # local desc.=500, Desc. dim.=1282021.03 | 82.7 | — | |
| RRT# Reranked images=100, Desc. version=R50-v12021.03 | 82.7 | — | |
| RRT# local desc.=500, # Reranked images=100, Desc. version=R50-v12021.03 | 82.7 | — | |
| GV*# local desc.=1000, # Reranked images=200, Desc. version=R50-v12021.03 | 82.5 | — | |
| SuperGlueDesc. version=SuperPoint [17], # local desc.=1024, Desc. dim.=256, Global Retrieval Baseline=DELG R50-v12021.03 | 82.3 | — | |
| GV# local desc.=1000, # Reranked images=100, Desc. version=R50-v12021.03 | 82.3 | — | |
| SuperGlueDesc. version=SuperPoint [17], # local desc.=500, Desc. dim.=256, Global Retrieval Baseline=DELG R50-v12021.03 | 81.9 | — | |
| DELG globalDesc. version=R50-v1, # local desc.=02021.03 | 81.6 | — | |
| DELG globalDesc. version=R50-v12021.03 | 81.6 | — | |
| DELG global# local desc.=0, # Reranked images=0, Desc. version=R50-v12021.03 | 81.6 | — | |
| HOW N/DBackbone=ResNet-18, Paired day-night training=true2023.09 | 81.4 | — | |
| CycleGAN CDBackbone=ResNet-18, Paired day-night training=false2023.09 | 79.8 | — | |
| HEDNGAN CDBackbone=ResNet-18, Paired day-night training=false2023.09 | 79.7 | — | |
| CiDeRTrain Set=SfM-120k, Backbone=ResNet-101, Pooling=GeM, Loss=ArcFace, Fine-tuned=false, Dimension=20482024.04 | 79.4 | — | |
| HOWBackbone=ResNet-18, Paired day-night training=false2023.09 | 79.4 | — | |
| RCFNGAN CDBackbone=ResNet-18, Paired day-night training=false2023.09 | 79.4 | — | |
| R-ASMK*Train Set=NC-clean, Backbone=ResNet-50, Pooling=Softmax, Local, Fine-tuned=true2024.04 | 78.7 | — | |
| AGEMTrain Set=SfM-120k, Backbone=ResNet-101, Pooling=GeM, Loss=Siamese, Fine-tuned=true, Dimension=20482024.04 | 78.1 | — | |
| R-ASMK*+SPTrain Set=NC-clean, Backbone=ResNet-50, Pooling=Softmax, Local, Fine-tuned=true2024.04 | 78 | — | |
| GLAMTrain Set=SfM-120k, Backbone=ResNet-101, Pooling=GeM, Loss=ArcFace, Fine-tuned=true, Dimension=5122024.04 | 77.5 | — | |
| Radenovic et al.Train Set=SfM-120k, Backbone=ResNet-101, Pooling=GeM, Loss=Siamese, Fine-tuned=true, Dimension=20482024.04 | 77.2 | — | |
| DELF-ASMK*+SPTrain Set=SfM-120k, Backbone=ResNet-50, Pooling=Softmax, Fine-tuned=false2024.04 | 76.9 | — | |
| DINOv1-S/16Train Set=GLDv2, Method=PS2025.04 | 75.3 | — | |
| Supervisedep=100, Backbone=ResNet-502023.10 | 74 | — | |
| RN101+R-MACTrain Set=ImNet-1k, Method=PS2025.04 | 74 | — | |
| SOLARTrain Set=SfM-120k, Backbone=ResNet-101, Pooling=GeM, Loss=Triplet, SOS, Fine-tuned=true, Dimension=20482024.04 | 70.9 | — | |
| DIRTrain Set=SfM-120k, Backbone=ResNet-101, Pooling=RMAC, Loss=Triplet, Fine-tuned=true, Dimension=20482024.04 | 68.3 | — | |
| DINOv1-S/16Train Set=ImNet-1k, Method=PS2025.04 | 63.1 | — | |
| HesAff-rSIFT-ASMK*+SPTrain Set=SfM-120k, Backbone=ResNet-50, Fine-tuned=false2024.04 | 61.4 | — | |
| MoCo-v3ep=1000, Backbone=ResNet-502023.10 | 59.1 | — | |
| CARPep=200, Backbone=ResNet-502023.10 | 58.8 | — | |
| Tripletep=980, Backbone=ResNet-502023.10 | 58.2 | — | |
| HgVT-STrain Set=ImNet-1k, Method=VS2025.04 | 56.7 | — | |
| DOLGTrain Set=SfM-120k, Backbone=ResNet-101, Pooling=GeM,GAP, Loss=ArcFace, Fine-tuned=true, Dimension=5122024.04 | 56.6 | — | |
| DINO (mc)ep=800, Backbone=ResNet-502023.10 | 55.9 | — | |
| HgVT-TiTrain Set=ImNet-1k, Method=VS2025.04 | 55.4 | — | |
| GroCoEpochs=400, Batch Size=1024, Views=2x224, Backbone=ResNet-502023.01 | 54.95 | — | |
| ALEX+GeMTrain Set=ImNet-1k, Method=PS2025.04 | 52.7 | — | |
| SimCLREpochs=400, Batch Size=4096, Views=2x224, Backbone=ResNet-502023.01 | 46.93 | — | |
| SimSiamEpochs=100, Batch Size=256, Views=2x224, Backbone=ResNet-502023.01 | 46.92 | — | |
| SwAVEpochs=400, Batch Size=4096, Views=2x224, Backbone=ResNet-502023.01 | 46.23 | — | |
| MoCo v2Epochs=200, Batch Size=256, Views=2x224, Backbone=ResNet-502023.01 | 42.8 | — | |
| RandomBackbone=ResNet-502023.10 | 4.1 | — |