Text-to-image retrieval on CLEVR-DS (test)
22.56Recall@1CLIP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CLIPSingle Feature=False, Input=CC@102022.08 | 22.56 | 45.45 | 64.91 | |
| DetailCLIPSingle Feature=True, Input=CC@102022.08 | 22.54 | 46.16 | 64.98 | |
| CLIPSingle Feature=True, Input=Full Image2022.08 | 10.61 | 29.64 | 49.5 |