Document Image Retrieval on NL-DIR (test)
36.17Recall@1SigLIP-So400m
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SigLIP-So400mEvaluation Protocol=Zero-shot, Model Category=Contrastive VLM2025.12 | 36.17 | 61.18 | 43.78 | |
| DFNEvaluation Protocol=Zero-shot, Model Category=Contrastive VLM2025.12 | 28.48 | 51.24 | 35.36 | |
| InternVL-14B-224pxEvaluation Protocol=Zero-shot, Model Category=Contrastive VLM2025.12 | 24.25 | 43.45 | 29.92 | |
| BLIP-large-COCOEvaluation Protocol=Zero-shot, Model Category=Contrastive VLM2025.12 | 5.95 | 13.8 | 8.14 | |
| BLIP-large-384Evaluation Protocol=Zero-shot, Model Category=Contrastive VLM2025.12 | 3.84 | 10.68 | 5.66 | |
| BLIP-baseEvaluation Protocol=Zero-shot, Model Category=Contrastive VLM2025.12 | 2.54 | 6.02 | 3.48 | |
| CLIP-baseEvaluation Protocol=Zero-shot, Model Category=Contrastive VLM2025.12 | 1.44 | 3.99 | 2.11 | |
| Qwen2-VLEvaluation Protocol=Zero-shot, Model Category=Generative VDU2025.12 | 0.29 | 1.66 | 0.59 | |
| UReaderEvaluation Protocol=Zero-shot, Model Category=Generative VDU2025.12 | 0.18 | 1.19 | 0.41 | |
| DocOwl1.5Evaluation Protocol=Zero-shot, Model Category=Generative VDU2025.12 | 0.1 | 0.94 | 0.29 | |
| DonutEvaluation Protocol=Zero-shot, Model Category=Generative VDU2025.12 | 0.02 | 0.21 | 0.07 | |
| Pix2StructEvaluation Protocol=Zero-shot, Model Category=Generative VDU2025.12 | 0.02 | 0.21 | 0.07 | |
| TextMonkeyEvaluation Protocol=Zero-shot, Model Category=Generative VDU2025.12 | 0.02 | 0.22 | 0.07 | |
| NougatEvaluation Protocol=Zero-shot, Model Category=Generative VDU2025.12 | 0.01 | 0.23 | 0.07 | |
| VaryEvaluation Protocol=Zero-shot, Model Category=Generative VDU2025.12 | 0.01 | 0.27 | 0.06 |