Composed Image Retrieval on CIRR Index Set v1.0
35.3Recall@1LinCIR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LinCIRParameters=2.6B, Backbone=CLIP-G2024.03 | 35.3 | 64.7 | 76.1 | — | |
| SEARLEParameters=2.6B, Backbone=CLIP-G2024.03 | 34.8 | 64.1 | 75.1 | — | |
| CIREVLParameters=14.6B*, Backbone=CLIP-G2024.03 | 34.7 | 64.3 | 75.1 | 91.7 | |
| SEARLEParameters=1.0B, Backbone=CLIP-H2024.03 | 34 | 64 | 75.3 | — | |
| LinCIRParameters=1.0B, Backbone=CLIP-H2024.03 | 33.8 | 63.5 | 73.4 | — | |
| MagicLens-LParameters=613M, Backbone=CoCa-L2024.03 | 33.3 | 67 | 77.9 | 94.4 | |
| Pic2WordParameters=987M, Backbone=CLIP-H2024.03 | 32.9 | 63.1 | 73.9 | — | |
| MagicLens-BParameters=267M, Backbone=CoCa-B2024.03 | 31.6 | 64 | 76.9 | 93.8 | |
| Pic2WordParameters=2.5B, Backbone=CLIP-G2024.03 | 30.4 | 58.1 | 69.2 | — | |
| MagicLens-LParameters=465M, Backbone=CLIP-L2024.03 | 30.1 | 61.7 | 74.4 | 92.6 | |
| PLIParameters=224M†, Backbone=BLIP-B2024.03 | 27.2 | 58.9 | 71.4 | 91.3 | |
| MagicLens-BParameters=166M, Backbone=CLIP-B2024.03 | 27 | 58 | 70.9 | 91.1 | |
| CompoDiffParameters=2.9B, Backbone=CLIP-G2024.03 | 26.7 | 55.1 | 74.5 | 92 | |
| Context-I2WParameters=496M, Backbone=CLIP-L2024.03 | 25.6 | 55.1 | 68.5 | 89.8 | |
| PLIParameters=428M†, Backbone=CLIP-L2024.03 | 25.5 | 54.6 | 67.6 | 88.7 | |
| LinCIRParameters=442M, Backbone=CLIP-L2024.03 | 25 | 53.3 | 66.7 | — | |
| CIREVLParameters=12.5B*, Backbone=CLIP-L2024.03 | 24.6 | 52.3 | 64.9 | 86.3 | |
| SEARLEParameters=442M, Backbone=CLIP-L2024.03 | 24.2 | 52.5 | 66.3 | 88.8 | |
| SEARLEParameters=165M, Backbone=CLIP-B2024.03 | 24 | 53.4 | 66.8 | 89.8 | |
| CIREVLParameters=12.3B*, Backbone=CLIP-B2024.03 | 23.9 | 52.5 | 66 | 87 | |
| Pic2WordParameters=429M, Backbone=CLIP-L2024.03 | 23.9 | 51.7 | 65.3 | 87.8 | |
| CompoDiffParameters=568M, Backbone=CLIP-L2024.03 | 18.2 | 53.1 | 70.8 | 90.3 | |
| PALAVRAParameters=176M, Backbone=CLIP-B2024.03 | 16.6 | 43.5 | 58.5 | 84 |