Visual Entity Recognition on InfoSeek
62.7Recall@1WikiCLIP-L
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| WikiCLIP-LScale=Large2026.03 | 62.7 | 77.7 | 82.2 | 86.6 | |
| WikiCLIP-SScale=Small2026.03 | 61.2 | 76.8 | 81.8 | 86.4 | |
| Echosight2026.03 | 53.2 | 74 | 77.4 | 77.9 | |
| CLIP I-IRetrieval Mode=Image-to-Image2026.03 | 45.6 | 67.1 | 73 | 77.9 | |
| CLIP I-TRetrieval Mode=Image-to-Text2026.03 | 32 | 54 | 61.6 | 68.2 |