Text-to-Video Retrieval on Something-to-Something v2
55.2Recall@1HiTeA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HiTeA# PT Data=5M2022.12 | 55.2 | 81.4 | 89.1 | |
| Singularity# PT Data=5M2022.12 | 44.1 | 73.5 | 82.2 | |
| Clip4Clip# PT Data=400M2022.12 | 43.1 | 71.4 | 80.7 | |
| PAXIONPatcher Training Loss=VTC+DVDM2023.05 | 28 | 54.3 | — | |
| KP-Perceiver FTPatcher Training Loss=VTC2023.05 | 27.1 | 56.8 | — | |
| Side-TuningPatcher Training Loss=VTC+DVDM2023.05 | 26.6 | 53.1 | — | |
| KP-Transformer FTPatcher Training Loss=VTC2023.05 | 21.7 | 46 | — | |
| InternVideo Backbone2023.05 | 19.9 | 40 | — |