Image Retrieval on INSTRE
83.1mAPIME layer
Evaluation results
| Method | Links | ||
|---|---|---|---|
| IME layerDimensions=512, Method category=Dimension reduction and manifold learning2017.07 | 83.1 | — | |
| IME layerDimensions=1024, Method category=Dimension reduction and manifold learning2017.07 | 82.8 | — | |
| IME layerDimensions=2048, Method category=Dimension reduction and manifold learning2017.07 | 82.4 | — | |
| IME layerDimensions=2048, Method category=Search reranking2017.07 | 82.4 | — | |
| ELViSTraining Domain=household items domain (SOP), Backbone=DINOv2, Descriptor Category=similarity-based2026.03 | 81.1 | — | |
| DiffusionDimensions=2048, Method category=Search reranking2017.07 | 80.5 | — | |
| ELViSTraining Domain=landmarks domain (GLDv2), Backbone=DINOv2, Descriptor Category=similarity-based2026.03 | 80.4 | — | |
| R2FormerTraining Domain=landmarks domain (GLDv2), Backbone=DINOv2, Descriptor Category=similarity-based2026.03 | 77.6 | — | |
| Chamfer+OTTraining Domain=landmarks domain (GLDv2), Backbone=DINOv2, Descriptor Category=similarity-based2026.03 | 76 | — | |
| Chamfer+OTTraining Domain=household items domain (SOP), Backbone=DINOv2, Descriptor Category=similarity-based2026.03 | 75.9 | — | |
| AMESTraining Domain=landmarks domain (GLDv2), Backbone=DINOv2, Descriptor Category=descriptor-based2026.03 | 75.6 | — | |
| R2FormerTraining Domain=household items domain (SOP), Backbone=DINOv2, Descriptor Category=similarity-based2026.03 | 73.4 | — | |
| RRTTraining Domain=landmarks domain (GLDv2), Backbone=DINOv2, Descriptor Category=descriptor-based2026.03 | 72.4 | — | |
| AMESTraining Domain=household items domain (SOP), Backbone=DINOv2, Descriptor Category=descriptor-based2026.03 | 72.2 | — | |
| SCSMDimensions=2048, Method category=Search reranking2017.07 | 71.4 | — | |
| QEDimensions=2048, Method category=Search reranking2017.07 | 70.5 | — | |
| IsoMapDimensions=512, Method category=Dimension reduction and manifold learning2017.07 | 69.7 | — | |
| IsoMapDimensions=1024, Method category=Dimension reduction and manifold learning2017.07 | 69.1 | — | |
| IsoMapDimensions=2048, Method category=Dimension reduction and manifold learning2017.07 | 68.6 | — | |
| RRTTraining Domain=household items domain (SOP), Backbone=DINOv2, Descriptor Category=descriptor-based2026.03 | 68.4 | — | |
| No re-rankingTraining Domain=landmarks domain (GLDv2), Backbone=DINOv22026.03 | 65.3 | — | |
| No re-rankingTraining Domain=household items domain (SOP), Backbone=DINOv22026.03 | 65.3 | — | |
| ChamferTraining Domain=household items domain (SOP), Backbone=DINOv2, Descriptor Category=similarity-based2026.03 | 63.2 | — | |
| R-MACDimensions=2048, Method category=Original representations2017.07 | 62.6 | — | |
| PCADimensions=2048, Method category=Dimension reduction and manifold learning2017.07 | 62.2 | — | |
| ICADimensions=2048, Method category=Dimension reduction and manifold learning2017.07 | 62.2 | — | |
| LLEDimensions=512, Method category=Dimension reduction and manifold learning2017.07 | 60.2 | — | |
| PCADimensions=1024, Method category=Dimension reduction and manifold learning2017.07 | 58.7 | — | |
| ICADimensions=1024, Method category=Dimension reduction and manifold learning2017.07 | 58.7 | — | |
| ChamferTraining Domain=landmarks domain (GLDv2), Backbone=DINOv2, Descriptor Category=similarity-based2026.03 | 55 | — | |
| LLEDimensions=1024, Method category=Dimension reduction and manifold learning2017.07 | 50.4 | — | |
| ICADimensions=512, Method category=Dimension reduction and manifold learning2017.07 | 50.3 | — | |
| PCADimensions=512, Method category=Dimension reduction and manifold learning2017.07 | 50 | — | |
| R-MACDimensions=512, Method category=Original representations2017.07 | 47.7 | — | |
| LLEDimensions=2048, Method category=Dimension reduction and manifold learning2017.07 | 42.7 | — | |
| AMESGlobal Descriptor=PE [7] with linear adaptation [26], Prompting Strategy=generic prompt, Local Descriptors=400 (102kB)2026.04 | — | 91.9 | |
| AMESGlobal Descriptor=DINOv3 [49], Prompting Strategy=generic prompt, Local Descriptors=400 (102kB)2026.04 | — | 85.1 | |
| LamRAGlobal Descriptor=PE [7] with linear adaptation [26], Prompting Strategy=generic prompt2026.04 | — | 95.3 | |
| LamRAGlobal Descriptor=PE [7] with linear adaptation [26], Prompting Strategy=task-specific prompt2026.04 | — | 96.6 | |
| LamRAGlobal Descriptor=DINOv3 [49], Prompting Strategy=generic prompt2026.04 | — | 89.1 | |
| LamRAGlobal Descriptor=DINOv3 [49], Prompting Strategy=task-specific prompt2026.04 | — | 90.5 | |
| No re-rankGlobal Descriptor=PE [7] with linear adaptation [26], Prompting Strategy=N/A2026.04 | — | 92.6 | |
| No re-rankGlobal Descriptor=DINOv3 [49], Prompting Strategy=N/A2026.04 | — | 82.2 | |
| QwenGlobal Descriptor=PE [7] with linear adaptation [26], Prompting Strategy=generic prompt2026.04 | — | 94.6 | |
| QwenGlobal Descriptor=PE [7] with linear adaptation [26], Prompting Strategy=task-specific prompt2026.04 | — | 96.4 | |
| QwenGlobal Descriptor=DINOv3 [49], Prompting Strategy=generic prompt2026.04 | — | 89.1 | |
| QwenGlobal Descriptor=DINOv3 [49], Prompting Strategy=task-specific prompt2026.04 | — | 90.7 | |
| Qwen-CGlobal Descriptor=PE [7] with linear adaptation [26], Prompting Strategy=task-specific prompt2026.04 | — | 95.2 | |
| Qwen-CGlobal Descriptor=DINOv3 [49], Prompting Strategy=task-specific prompt2026.04 | — | 89.7 |