Composed Image Retrieval on FashionIQ (Toptee subset)
50.48Recall@10LinCIR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LinCIRArch=ViT-G/14, Train=true2026.05 | 50.48 | 71.09 | |
| OSCAR2026.02 | 48.24 | 71.24 | |
| RzenEmbed-v2-7BParameters=7B2026.02 | 46.56 | 67.57 | |
| QQMM-embed-v2Version=v22026.02 | 46.35 | 68.49 | |
| SEIZEArch=ViT-G/14, Train=false2026.05 | 45.94 | 71.12 | |
| X^R2026.02 | 43.91 | 62.57 | |
| STiTchArch=ViT-G/14, Train=false2026.05 | 42.86 | 64.95 | |
| MRA-CIR2026.02 | 41.25 | 62.51 | |
| OSrCIRArch=ViT-G/14, Train=false2026.05 | 41.04 | 61.83 | |
| SEARLEArch=ViT-G/14, Train=true2026.05 | 39.83 | 61.45 | |
| CIReVLArch=ViT-G/14, Train=false2026.05 | 35.8 | 56.14 | |
| FiRE2026.02 | 35.64 | 57.83 | |
| Pic2WordArch=ViT-G/14, Train=true2026.05 | 35.24 | 57.62 | |
| MagicLensBackbone=CLIP ViT-L/142026.07 | 34 | 57.7 | |
| FlowCIRBackbone=CLIP ViT-L/142026.07 | 33.2 | 53.9 | |
| AutoCIR2026.02 | 33.1 | 55.58 | |
| LDRE2026.02 | 31.57 | 53.64 | |
| CIReVL2026.02 | 31.36 | 53.65 | |
| SEARLE-XL-OTIScale=XL-OTI2026.02 | 30.9 | 51.76 | |
| Context-I2WBackbone=CLIP ViT-L/142026.07 | 30.6 | 52.9 | |
| MagicLensBackbone=CLIP ViT-B/322026.07 | 30.2 | 52.3 | |
| SEARLE2026.02 | 29.32 | 49.97 | |
| SEARLEBackbone=CLIP ViT-L/142026.07 | 29.3 | 50 | |
| LinCIR2026.02 | 28.81 | 50.18 | |
| LinCIRBackbone=CLIP ViT-L/142026.07 | 28.8 | 50.2 | |
| Pic2Word2026.02 | 27.9 | 47.4 | |
| Pic2WordBackbone=CLIP ViT-L/142026.07 | 27.9 | 47.4 | |
| Ops-MM-v1-7BParameters=7B2026.02 | 27.69 | 46.51 | |
| FlowCIRBackbone=CLIP ViT-B/322026.07 | 26.4 | 47.8 | |
| SEARLEBackbone=CLIP ViT-B/322026.07 | 25.6 | 46.2 | |
| Qwen3-Embed-8BParameters=8B2026.02 | 23.41 | 36.97 | |
| Qwen3-Embed-4BParameters=4B2026.02 | 21.88 | 36.05 | |
| PALAVRABackbone=CLIP ViT-B/322026.07 | 20.6 | 38.8 | |
| Text-onlyBackbone=CLIP ViT-B/322026.07 | 20.2 | 40.5 | |
| Text-onlyBackbone=CLIP ViT-L/142026.07 | 17.4 | 33.9 | |
| bge-m32026.02 | 15.96 | 27.33 | |
| Qwen3-Embed-0.6BParameters=0.6B2026.02 | 15.96 | 27.84 | |
| B3_Qwen2_7BBackbone=Qwen2, Parameters=7B2026.02 | 14.99 | 29.88 | |
| VLM2Vec2026.02 | 11.37 | 22.95 | |
| Image-onlyBackbone=CLIP ViT-L/142026.07 | 8.4 | 16.5 | |
| Image-onlyBackbone=CLIP ViT-B/322026.07 | 6.2 | 13.4 |