Composed Image Retrieval on CIRR Subset (test)
78.68R@1CASE (Text-only)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CASE (Text-only)Mode=Modality Redundancy2023.03 | 78.68 | 91.7 | 96.08 | |
| CASE Pre-LaSCo.Ca.Mode=Pre-Train2023.03 | 76.48 | 90.37 | 95.71 | |
| CASE Pre-LaSCoMode=Pre-Train2023.03 | 76.39 | 90.12 | 95.86 | |
| CASEMode=Train2023.03 | 75.88 | 90.58 | 96 | |
| Pic2WordBackbone=ViT-G2023.12 | 68.92 | 85.45 | 93.04 | |
| SEARLEBackbone=ViT-G2023.12 | 68.72 | 84.7 | 93.23 | |
| SEARLEBackbone=ViT-H2023.12 | 64.63 | 83.21 | 92.77 | |
| CASE - LaSCo.Ca.Mode=Zero Shot2023.03 | 64.29 | 82.66 | 91.61 | |
| LinCIRBackbone=ViT-G2023.12 | 63.35 | 82.22 | 91.98 | |
| LinCIRBackbone=ViT-H2023.12 | 62.43 | 81.47 | 92.12 | |
| LF-CLIPMode=Train2023.03 | 62.39 | 81.81 | 92.02 | |
| Pic2WordBackbone=ViT-H2023.12 | 62.22 | 81.35 | 91.23 | |
| CASE - LaSCoMode=Zero Shot2023.03 | 60.17 | 80.17 | 90.41 | |
| Ours (m=0)Training Setting=Zero-shot, Interaction Turns=m=02026.05 | 58.4 | 80.1 | 89.7 | |
| CLARASetting=Zero-shot, Turns=m=02026.06 | 57.2 | 78.6 | 89.4 | |
| LinCIRBackbone=ViT-L2023.12 | 57.11 | 77.37 | 88.89 | |
| OSrCIRTraining Setting=Zero-shot, Interaction Turns=m=02026.05 | 56.9 | 79 | 89.2 | |
| OSrCIRSetting=Zero-shot, Turns=m=02026.06 | 56.9 | 79 | 89.2 | |
| CASE Init.Mode=Zero Shot2023.03 | 55.74 | 77.1 | 88.48 | |
| CompoDiffTraining Setting=Zero-shot, Interaction Turns=m=02026.05 | 55 | 77.6 | 88.6 | |
| CompoDiffSetting=Zero-shot, Turns=m=02026.06 | 55 | 77.6 | 88.6 | |
| SEARLETraining Setting=Zero-shot, Interaction Turns=m=02026.05 | 53.8 | 76.9 | 88.1 | |
| SEARLESetting=Zero-shot, Turns=m=02026.06 | 53.8 | 76.9 | 88.1 | |
| Pic2WordBackbone=ViT-L2023.12 | 53.76 | 74.46 | 87.08 | |
| SEARLEBackbone=ViT-L2023.12 | 53.76 | 75.01 | 88.19 | |
| Pic2WordTraining Setting=Zero-shot, Interaction Turns=m=02026.05 | 51.1 | 74.4 | 87 | |
| Pic2WordSetting=Zero-shot, Turns=m=02026.06 | 51.1 | 74.4 | 87 | |
| LF-BLIPMode=Train2023.03 | 50.22 | 73.16 | 86.82 | |
| ARTEMISMode=Train2023.03 | 39.99 | 62.2 | 75.67 | |
| CIRPLANTMode=Train2023.03 | 39.2 | 63.03 | 79.49 | |
| CIRPLANTTraining Setting=Supervised, Interaction Turns=m=02026.05 | 39.2 | 63 | 79.5 | |
| CIRPLANTSetting=Sup., Turns=m=02026.06 | 39.2 | 63 | 79.5 | |
| TIRGTraining Setting=Supervised, Interaction Turns=m=02026.05 | 22.7 | 45 | 65.1 | |
| TIRGSetting=Sup., Turns=m=02026.06 | 22.7 | 45 | 65.1 | |
| TIRGMode=Train2023.03 | 22.67 | 44.97 | 65.14 | |
| MAAF-BERTMode=Train2023.03 | 22.04 | 42.41 | 62.14 | |
| MAAF-RPMode=Train2023.03 | 21.41 | 42.17 | 61.6 | |
| MAAFMode=Train2023.03 | 21.05 | 41.81 | 61.6 | |
| CASE (Image-only)Mode=Modality Redundancy2023.03 | 19.78 | 39.49 | 59.87 | |
| RandomMode=Train2023.03 | 16.67 | 33.33 | 50 |