Text-to-image retrieval on CLEVR-DS-L (test)
16.33Recall@1DetailCLIP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DetailCLIPSingle Feature=True, Input=CC@102022.08 | 16.33 | 25.47 | 30.96 | |
| CLIPSingle Feature=False, Input=CC@102022.08 | 15.3 | 25.93 | 31.97 | |
| CLIPSingle Feature=True, Input=Full Image2022.08 | 13.57 | 23.19 | 28.65 |