Image-Text Retrieval on Flickr30K (IR@1, TR@1, Avg.)
41.5Image Retrieval Recall@1CLIP + MM-TS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CLIP + MM-TSBackbone=RN50, Evaluation Protocol=zero-shot, Pre-training Dataset=CC3M2026.03 | 41.5 | 54.3 | 47.9 | |
| CLIPBackbone=RN50, Evaluation Protocol=zero-shot, Pre-training Dataset=CC3M2026.03 | 40.9 | 50.9 | 45.9 |