Image-to-Text Retrieval on Fashion200K
34.9R@5CLIP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CLIPvariant=oracle2026.04 | 34.9 | — | — | |
| ToMA-domain2026.04 | 23.8 | — | — | |
| ToMA2026.04 | 23.5 | — | — | |
| SemiCLIP2026.04 | 23.3 | — | — | |
| MOON2.0mode=Zero-shot2025.11 | 23.16 | 27.09 | 13.1 | |
| GMEmode=Zero-shot2025.11 | 21.14 | 27.23 | 10.31 | |
| MOONmode=Zero-shot2025.11 | 20.02 | 25.09 | 11.71 | |
| S-CLIP2026.04 | 16.9 | — | — | |
| MM-Embedmode=Zero-shot2025.11 | 15.91 | 22.57 | 6.12 | |
| Soft-PL2026.04 | 15.2 | — | — | |
| BGE-VL-Largemode=Zero-shot2025.11 | 14.21 | 18.65 | 7.27 | |
| CLIPvariant=fine-tuned2026.04 | 13.7 | — | — | |
| Hard-PL2026.04 | 12.6 | — | — | |
| CLIPvariant=original2026.04 | 12.5 | — | — | |
| SigLIP2mode=Zero-shot2025.11 | 10.49 | 14.7 | 4.39 | |
| CASLIE-Smode=Zero-shot2025.11 | 10.04 | 13.89 | 4.41 | |
| FashionCLIPmode=Zero-shot2025.11 | 9.15 | 12.97 | 4.34 | |
| Qwen2.5-VL-3Bmode=Zero-shot2025.11 | 1.17 | 1.63 | 0.54 | |
| InternVL3-2Bmode=Zero-shot2025.11 | 0.11 | 0.62 | 0.12 | |
| BLIP-FFRetrieval Strategy=Embedding-based, Fusion Strategy=Feature-level Fusion, Re-ranking=false, Retrieval Pool=Task-specific2025.03 | — | 28.9 | — | |
| BLIP-FFApproach=Embedding-based, Fusion=FF2025.03 | — | 27.6 | — | |
| CLIPversion=original2026.04 | — | — | 4.7 | |
| CLIPversion=fine-tuned2026.04 | — | — | 4.8 | |
| CLIP-SFRetrieval Strategy=Embedding-based, Fusion Strategy=Score-level Fusion, Re-ranking=false, Retrieval Pool=Task-specific2025.03 | — | 18.3 | — | |
| CLIP-SFApproach=Embedding-based, Fusion=SF2025.03 | — | 17.9 | — | |
| GENIUSRetrieval Strategy=Generative, Fusion Strategy=None, Re-ranking=false, Retrieval Pool=Task-specific2025.03 | — | 12.8 | — | |
| GENIUSRetrieval Strategy=Generative, Fusion Strategy=None, Re-ranking=true, Retrieval Pool=Task-specific2025.03 | — | 16.3 | — | |
| GENIUSApproach=Generative2025.03 | — | 12.8 | — | |
| GENIUSApproach=Generative, Re-ranking=true2025.03 | — | 16.2 | — | |
| Hard-PL2026.04 | — | — | 4.8 | |
| S-CLIP2026.04 | — | — | 6.6 | |
| SemiCLIP2026.04 | — | — | 10.1 | |
| Soft-PL2026.04 | — | — | 5.6 | |
| ToMA2026.04 | — | — | 10 | |
| ToMA-domain2026.04 | — | — | 9.9 |