Audio-Visual Recognition on Kinetics-Sounds (val)
90.9Top-1 Acc (KSounds)CAV-MAE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CAV-MAEOut-of-Domain Pretrain=ImageNet + AudioSet SSL, In-Domain Training=K400 SSL + SL2022.10 | 90.9 | 85 | |
| CAV-MAEOut-of-Domain Pretrain=ImageNet SSL, In-Domain Training=K400 SSL + SL2022.10 | 90.6 | 83.2 | |
| MBTOut-of-Domain Pretrain=ImageNet SL, In-Domain Training=K400 SL2022.10 | 85 | — | |
| CAV-MAEOut-of-Domain Pretrain=No, In-Domain Training=K400 SSL + SL2022.10 | 83.3 | 70.6 |