Text-to-audiovisual Retrieval on VALOR-32K (test)
80.9Recall@1VALOR_L
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| VALOR_L#Example=33.5M, Mod=V+A, Dual Softmax (+DSL)=true2023.04 | 80.9 | 93.9 | 97.1 | |
| VALOR_L#Example=33.5M, Mod=V+A, Dual Softmax (+DSL)=false2023.04 | 73.2 | 91.6 | 95.4 | |
| VALOR_B#Example=6.5M, Mod=V+A2023.04 | 67.9 | 89.7 | 94.4 | |
| CLIP4ClipMod=V2023.04 | 43.4 | 69.9 | 79.7 | |
| VALOR_E#Example=5.5M, Mod=V2023.04 | 43.3 | 70.3 | 80 | |
| Frozen#Example=6.1M, Mod=V2023.04 | 32.9 | 60.4 | 71.2 | |
| AVLNet#Example=136M, Mod=V+A2023.04 | 21.6 | 42.2 | 53.5 |