Image-to-Text Retrieval on Flickr
98.8R@1LamRA
Evaluation Results
| Method | Links | |
|---|---|---|
| LamRAReranker=true2025.07 | 98.8 | |
| U-MARVEL+Reranker=true2025.07 | 97.7 | |
| LamRAreranking=pointwise/listwise2024.12 | 97.6 | |
| EVA-CLIPparameters=18B2024.12 | 96.7 | |
| U-MARVELReranker=false2025.07 | 96.7 | |
| PE-coreArchitecture=G/142026.04 | 96.2 | |
| UniMEReranker=false2025.07 | 95.9 | |
| EVA-CLIPparameters=8B2024.12 | 95.6 | |
| EVA-CLIP-8BSetting=Zero-shot2026.02 | 95.6 | |
| SigLIP2Architecture=g/162026.04 | 95.4 | |
| LamRA-RetReranker=false2025.07 | 95.1 | |
| TIPSv2Architecture=g/142026.04 | 95.1 | |
| SigLIP 2Setting=Zero-shot2026.02 | 94.9 | |
| UniIR-CLIP2024.12 | 94.2 | |
| SigLIPSetting=Zero-shot2026.02 | 93.5 | |
| TIPSArchitecture=g/142026.04 | 93 | |
| OpenCLIPArchitecture=G/142026.04 | 92.9 | |
| LamRA-Ret2024.12 | 92.7 | |
| VersaViTVision backbone state=Unfrozen, Projection layer fine-tuning=true, Training data alignment=30M image-text pairs2026.02 | 92.5 | |
| VLM2VecReranker=false2025.07 | 90.8 | |
| Long-CLIP-L2024.12 | 89.3 | |
| VersaViTVision backbone state=Frozen, Projection layer fine-tuning=true, Training data alignment=30M image-text pairs2026.02 | 88.3 | |
| E5-V2024.12 | 88.2 | |
| CLIP-L2024.12 | 87.2 | |
| CLIP-LSetting=Zero-shot2026.02 | 85.2 | |
| CLIPArchitecture=L/142026.04 | 85.2 | |
| MagicLens-L2024.12 | 84.6 | |
| E5-VReranker=false2025.07 | 84 | |
| ViT-B/16Architecture=ViT-B/16, Zero-shot evaluation=true2026.03 | 76.5 | |
| ViT-B/16Backbone=ViT-B/16, Zero-shot=true, Role=Teacher2026.03 | 76.5 | |
| UniIR-CLIP_SFReranker=false2025.07 | 75 | |
| CLIP-KDBackbone=ResNet-50, Zero-shot=true2026.03 | 74.9 | |
| CLIP-RDBackbone=ResNet-50, Zero-shot=true2026.03 | 73 | |
| CLIP-RDBackbone=EfficientNet-B0, Zero-shot=true2026.03 | 64.3 | |
| MagicLens-LReranker=false2025.07 | 59.3 | |
| TinyCLIPArchitecture=TinyCLIP, Zero-shot evaluation=true2026.03 | 58.6 | |
| CLIP-KD*Architecture=ViT-T/16, Zero-shot evaluation=true2026.03 | 58.4 | |
| CLIP-RDArchitecture=ViT-T/16, Zero-shot evaluation=true2026.03 | 58.3 | |
| CLIP-KDBackbone=ResNet-18, Zero-shot=true2026.03 | 56.6 | |
| CLIP-RDBackbone=ResNet-18, Zero-shot=true2026.03 | 54.4 | |
| CLIP-LReranker=false2025.07 | 52.8 | |
| ResNet-50Backbone=ResNet-50, Zero-shot=true, Role=Student2026.03 | 52.2 | |
| EfficientNet-B0Backbone=EfficientNet-B0, Zero-shot=true, Role=Student2026.03 | 44.9 | |
| ResNet-18Backbone=ResNet-18, Zero-shot=true, Role=Student2026.03 | 43.6 | |
| ViT-T/16Architecture=ViT-T/16, Zero-shot evaluation=true2026.03 | 39.3 |