Image Retrieval on Revisited Oxford (ROxf) (Hard)
80.7mAPSuperGlobal
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SuperGlobalBackbone=ResNet-101, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=8002023.08 | 80.7 | — | |
| SuperGlobalBackbone=ResNet-101, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=16002023.08 | 80.6 | — | |
| SuperGlobalBackbone=ResNet-101, Retrieval Category=SuperGlobal retrieval and reranking, Reranking Setting=Rerank top 4002023.08 | 80.2 | — | |
| SuperGlobalBackbone=ResNet-101, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=4002023.08 | 80.2 | — | |
| SuperGlobalBackbone=ResNet-50, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=8002023.08 | 77.4 | — | |
| SuperGlobalBackbone=ResNet-50, Retrieval Category=SuperGlobal retrieval and reranking, Reranking Setting=Rerank top 4002023.08 | 77.1 | — | |
| SuperGlobalBackbone=ResNet-50, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=4002023.08 | 77.1 | — | |
| SuperGlobalBackbone=ResNet-50, Strategy=SuperGlobal feature retrieval and reranking, Reranking Depth=16002023.08 | 76.9 | — | |
| CVNetBackbone=ResNet-101, Retrieval Category=Global feature retrieval + Local feature reranking, Reranking Setting=Rerank top 4002023.08 | 75.9 | — | |
| CVNetBackbone=ResNet-101, Strategy=Global feature retrieval + Local feature reranking, Reranking Depth=4002023.08 | 75.9 | — | |
| CVNetBackbone=ResNet-50, Retrieval Category=Global feature retrieval + Local feature reranking, Reranking Setting=Rerank top 4002023.08 | 75.6 | — | |
| CVNetBackbone=ResNet-50, Strategy=Global feature retrieval + Local feature reranking, Reranking Depth=4002023.08 | 75.6 | — | |
| SuperGlobalBackbone=ResNet-101, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 72.1 | — | |
| SuperGlobalBackbone=ResNet-101, Strategy=Global feature retrieval2023.08 | 72.1 | — | |
| SuperGlobalBackbone=ResNet-50, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 67.7 | — | |
| SuperGlobalBackbone=ResNet-50, Strategy=Global feature retrieval2023.08 | 67.7 | — | |
| RRT# Reranked images=400, Desc. version=R50-v2-clean2021.03 | 64.2 | — | |
| alphaQE + RRT# Reranked images=200, Desc. version=R50-v2-clean2021.03 | 64 | — | |
| DELGBackbone=ResNet-101, Retrieval Category=Global feature retrieval + Local feature reranking, Reranking Setting=GV rerank top 1002023.08 | 64 | — | |
| DELGBackbone=ResNet-101, Strategy=Global feature retrieval + Local feature reranking, Reranking Depth=1002023.08 | 64 | — | |
| CVNetBackbone=ResNet-101, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 63.1 | — | |
| CVNetBackbone=ResNet-101, Strategy=Global feature retrieval2023.08 | 63.1 | — | |
| RRT# Reranked images=200, Desc. version=R50-v2-clean2021.03 | 62.5 | — | |
| RRT# local desc.=500, # Reranked images=200, Desc. version=R50-v2-clean2021.03 | 62.5 | — | |
| CVNetBackbone=ResNet-50, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 62.1 | — | |
| CVNetBackbone=ResNet-50, Strategy=Global feature retrieval2023.08 | 62.1 | — | |
| CiDeR-FTTraining Set=RGLDv2-clean, Fine-tuned=true2024.04 | 61.9 | 70.4 | |
| DOLGBackbone=ResNet-101, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 61.1 | — | |
| DOLGBackbone=ResNet-101, Strategy=Global feature retrieval2023.08 | 61.1 | — | |
| RRT# Reranked images=400, Desc. version=R50-v12021.03 | 60.5 | — | |
| RRT# Reranked images=100, Desc. version=R50-v2-clean2021.03 | 60.2 | — | |
| RRT# local desc.=500, # Reranked images=100, Desc. version=R50-v2-clean2021.03 | 60.2 | — | |
| alphaQE + RRT# Reranked images=200, Desc. version=R50-v12021.03 | 59.8 | — | |
| CiDeR-FTTraining Set=SfM-120k, Fine-tuned=true2024.04 | 58.9 | 71.1 | |
| RRT# Reranked images=200, Desc. version=R50-v12021.03 | 58.8 | — | |
| RRT# local desc.=500, # Reranked images=200, Desc. version=R50-v12021.03 | 58.8 | — | |
| DOLGBackbone=ResNet-50, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 58.8 | — | |
| DOLGBackbone=ResNet-50, Strategy=Global feature retrieval2023.08 | 58.8 | — | |
| GV# local desc.=1000, # Reranked images=100, Desc. version=R50-v2-clean2021.03 | 57.9 | — | |
| DELGBackbone=ResNet-50, Retrieval Category=Global feature retrieval + Local feature reranking, Reranking Setting=GV rerank top 1002023.08 | 57.9 | — | |
| DELGBackbone=ResNet-50, Strategy=Global feature retrieval + Local feature reranking, Reranking Depth=1002023.08 | 57.9 | — | |
| GV*# local desc.=1000, # Reranked images=200, Desc. version=R50-v2-clean2021.03 | 57.5 | — | |
| RRT# Reranked images=100, Desc. version=R50-v12021.03 | 56.4 | — | |
| RRT# local desc.=500, # Reranked images=100, Desc. version=R50-v12021.03 | 56.4 | — | |
| DELGBackbone=ResNet-101, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 55.6 | — | |
| DELGBackbone=ResNet-101, Strategy=Global feature retrieval2023.08 | 55.6 | — | |
| GV*# local desc.=1000, # Reranked images=200, Desc. version=R50-v12021.03 | 55.4 | — | |
| CiDeRTraining Set=RGLDv2-clean, Fine-tuned=false2024.04 | 54.9 | 66.6 | |
| alphaQEDesc. version=R50-v2-clean2021.03 | 54.6 | — | |
| GV# local desc.=1000, # Reranked images=100, Desc. version=R50-v12021.03 | 54.2 | — | |
| DELG globalDesc. version=R50-v2-clean2021.03 | 51 | — | |
| DELG global# local desc.=0, # Reranked images=0, Desc. version=R50-v2-clean2021.03 | 51 | — | |
| DELGBackbone=ResNet-50, Retrieval Category=Global feature retrieval, Reranking Setting=No reranking2023.08 | 51 | — | |
| DELGBackbone=ResNet-50, Strategy=Global feature retrieval2023.08 | 51 | — | |
| GLAMTraining Set=RGLDv2-clean, Fine-tuned=false2024.04 | 49.6 | 61.6 | |
| alphaQEDesc. version=R50-v12021.03 | 49.4 | — | |
| SOLARTraining Set=RGLDv2-clean, Fine-tuned=true2024.04 | 48 | 62.3 | |
| DOLGTraining Set=RGLDv2-clean, Fine-tuned=false2024.04 | 47.4 | 60 | |
| DELG globalDesc. version=R50-v12021.03 | 45.1 | — | |
| DELG global# local desc.=0, # Reranked images=0, Desc. version=R50-v12021.03 | 45.1 | — | |
| CiDeRTraining Set=SfM-120k, Fine-tuned=false2024.04 | 42.4 | 56.4 | |
| AGEMTraining Set=SfM-120k, Fine-tuned=true2024.04 | 40.7 | — | |
| GLAMTraining Set=SfM-120k, Fine-tuned=false2024.04 | 39.5 | — | |
| Filip et al.Training Set=SfM-120k, Fine-tuned=true2024.04 | 38.5 | 53 | |
| TokenTraining Set=RGLDv2-clean, Fine-tuned=false2024.04 | 37.3 | 54.1 | |
| Yokoo et al.Training Set=RGLDv2-clean, Fine-tuned=false2024.04 | 35.6 | 51.5 | |
| SOLARTraining Set=SfM-120k, Fine-tuned=true2024.04 | 27.1 | 41.4 | |
| GeMTraining Set=SfM-120k, Fine-tuned=false2024.04 | 25.8 | 42.2 | |
| DIRTraining Set=SfM-120k, Fine-tuned=false2024.04 | 25.5 | 42 | |
| DOLGTraining Set=SfM-120k, Fine-tuned=false2024.04 | 18.1 | 27.9 | |
| CARPep=200, Backbone=ResNet-502023.10 | 15.5 | — | |
| Tripletep=980, Backbone=ResNet-502023.10 | 12 | — | |
| DINO (mc)ep=800, Backbone=ResNet-502023.10 | 11.1 | — | |
| MoCo-v3ep=1000, Backbone=ResNet-502023.10 | 10.9 | — | |
| Supervisedep=100, Backbone=ResNet-502023.10 | 8.5 | — | |
| SwAVEpochs=400, Batch Size=4096, Views=2x224, Backbone=ResNet-502023.01 | 8.35 | — | |
| GroCoEpochs=400, Batch Size=1024, Views=2x224, Backbone=ResNet-502023.01 | 7.11 | — | |
| SimSiamEpochs=100, Batch Size=256, Views=2x224, Backbone=ResNet-502023.01 | 7.04 | — | |
| MoCo v2Epochs=200, Batch Size=256, Views=2x224, Backbone=ResNet-502023.01 | 5.07 | — | |
| SimCLREpochs=400, Batch Size=4096, Views=2x224, Backbone=ResNet-502023.01 | 4.56 | — | |
| RandomBackbone=ResNet-502023.10 | 0.7 | — |