Text-to-Video Retrieval on MSR-VTT 7K (train/test)
34.9R@1Query-scoring
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Query-scoringbackbone=CLIP, agg.=Q-score, # agg. params=12022.05 | 34.9 | 59.4 | 68.7 | 3 | 49.4 | |
| CAMOEbackbone=CLIP, agg.=S.E attn.2022.05 | 32.9 | 58.3 | 68.4 | 3 | 42.6 | |
| Clip2Videobackbone=CLIP, agg.=TDB, TAB, # agg. params=19M2022.05 | 29.8 | 55.5 | 66.2 | 4 | 45.4 | |
| TeachText2022.05 | 15 | 38.5 | 51.7 | 10 | — | |
| JSFusion2022.05 | 10.2 | 31.2 | 43.2 | 12 | — | |
| CE2022.05 | 10 | 29 | 41.2 | 16 | 86.8 |