Image-to-text Retrieval on MSCOCO (5K)
77.8R@1AltCLIP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| AltCLIPBackbone=ViT-L, Zero-shot=true, Stage=Contrastive Learning2022.11 | 77.8 | 94.4 | 97.5 | 85 | |
| AltCLIPTBackbone=ViT-L, Zero-shot=true, Stage=Teacher Learning2022.11 | 76.3 | 93.9 | 97.6 | 82.5 | |
| AMoEHead=Ensemble, Resolution=512x512, Parameters=0.3B-0.6B2025.12 | 72.14 | — | — | — | |
| CN-CLIPBackbone=ViT-L, Zero-shot=true2022.11 | 72.1 | 90.9 | 94.7 | 82.4 | |
| RADIOv2.5-HHead=Ensemble, Resolution=512x512, Parameters=0.6B2025.12 | 71.82 | — | — | — | |
| RADIOv2.5-LHead=Ensemble, Resolution=512x512, Parameters=0.3B2025.12 | 71.04 | — | — | — | |
| RADIOv2.5-HHead=CLIP, Resolution=512x512, Parameters=0.6B2025.12 | 70.92 | — | — | — | |
| RADIOv2.5-LHead=CLIP, Resolution=512x512, Parameters=0.3B2025.12 | 69.42 | — | — | — | |
| VISTA-LTime (ms)=~40, Fine-tuning=true2022.03 | 68.9 | 90.1 | 95.4 | — | |
| AMoEHead=SigLIP2, Resolution=512x512, Parameters=0.3B-0.6B2025.12 | 68.18 | — | — | — | |
| RADIOv2.5-HHead=SigLIP, Resolution=512x512, Parameters=0.6B2025.12 | 67.66 | — | — | — | |
| RADIOv2.5-LHead=SigLIP, Resolution=512x512, Parameters=0.3B2025.12 | 67.6 | — | — | — | |
| TeachersHead=SigLIP2, Resolution=512x5122025.12 | 67.42 | — | — | — | |
| VSEinftyFine-tuning=true2022.03 | 66.4 | 89.3 | — | — | |
| SOHOFine-tuning=true2022.03 | 66.4 | 88.2 | 93.8 | — | |
| AMoEHead=DINOv3, Resolution=512x512, Parameters=0.3B-0.6B2025.12 | 65.8 | — | — | — | |
| TeachersHead=DINOv3, Resolution=512x5122025.12 | 64.44 | — | — | — | |
| UNITER-BTime (ms)=~900, Fine-tuning=true2022.03 | 64.4 | 87.4 | 93.1 | — | |
| VISTA-BTime (ms)=~17, Fine-tuning=true2022.03 | 63.9 | 87.8 | 93.6 | — | |
| Pixel-BERT-XTime (ms)=~160, Fine-tuning=true2022.03 | 63.6 | 87.5 | 93.6 | — | |
| Unicoder-VLTime (ms)=925, Fine-tuning=true2022.03 | 62.3 | 87.1 | 92.8 | — | |
| ViLT-BTime (ms)=~15, Fine-tuning=true2022.03 | 61.5 | 86.3 | 92.7 | — | |
| Pixel-BERT-RTime (ms)=~60, Fine-tuning=true2022.03 | 59.8 | 85.5 | 91.6 | — | |
| TaiyiBackbone=ViT-L, Zero-shot=true2022.11 | 58.1 | 83.9 | 91.7 | 73.3 | |
| SGRAFFine-tuning=true2022.03 | 57.8 | — | 91.6 | — | |
| ViLT-B/32Visual Embed=Linear, Time (ms)=~15, Zero-Shot=true, Pre-training steps=200K2021.02 | 56.5 | 82.6 | 89.6 | — | |
| IMRAMFine-tuning=true2022.03 | 53.7 | 83.2 | 91 | — | |
| ViLT-B/32Visual Embed=Linear, Time (ms)=~15, Zero-Shot=true2021.02 | 53.4 | 80.7 | 88.8 | — | |
| VSRNFine-tuning=true2022.03 | 53 | 81.1 | 89.4 | — | |
| SCANFine-tuning=true2022.03 | 50.4 | 82.2 | 90 | — | |
| PCMEFine-tuning=true2022.03 | 44.2 | 73.8 | 83.6 | — | |
| ImageBERTVisual Embed=Region, Time (ms)=~925, Zero-Shot=true, Additional Pre-training Dataset=10M proprietary2021.02 | 44 | 71.2 | 80.4 | — | |
| UniConBackbone=CLIP ViT-B/32, Train time (s)=11.152026.04 | 32.9 | — | 68.5 | — | |
| SGD–CLIPBackbone=CLIP ViT-B/32, Train time (s)=1066.602026.04 | 12.8 | — | 41.5 | — | |
| UniConBackbone=RN-50 + SBERT, Train time (s)=11.112026.04 | 10.5 | — | 38.8 | — | |
| DINORANKCLIPBackbone=ViT-B/32, Pre-training Dataset=CC3M, R=32026.05 | 8.05 | 21.2 | 30.4 | — | |
| RANKCLIPBackbone=ViT-B/32, Pre-training Dataset=CC3M2026.05 | 7.18 | 19.46 | 28.48 | — | |
| CLIPBackbone=ViT-B/32, Pre-training Dataset=CC3M2026.05 | 6.68 | 18.36 | 26.94 | — | |
| CyCLIPBackbone=ViT-B/32, Pre-training Dataset=CC3M2026.05 | 6.5 | 19.34 | 29.14 | — | |
| ALIPBackbone=ViT-B/32, Pre-training Dataset=CC3M2026.05 | 6.04 | 18.04 | 26.92 | — | |
| SGD–CLIPBackbone=RN-50 + SBERT, Train time (s)=5121.722026.04 | 5.3 | — | 25.3 | — | |
| CLIPBackbone=ViT-L, Zero-shot=true2022.11 | 3.5 | 8.9 | 14.4 | 6.2 |