Audio-visual event classification on AudioSet 2M
49.1mAP (Audio-only)EquiAV
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| EquiAVPretrain=SSL2024.03 | 49.1 | 30.1 | 54.6 | |
| MAVILPretrain=SSL2024.03 | 48.7 | 30.3 | 53.3 | |
| MAVIL2024.09 | 48.7 | 30.3 | 53.3 | |
| CAV-MAEPretrain=SSL2024.03 | 46.6 | 26.2 | 51.2 | |
| AudiovisualMAE*Pretrain=SSL2024.03 | 46.6 | 31.1 | 51.8 | |
| CAV-MAE2024.09 | 46.6 | 26.2 | 51.2 | |
| MBT*Pretrain=IN21K SL2024.03 | 41.5 | 31.3 | 49.6 | |
| MBT2024.09 | 41.5 | 31.3 | 49.6 | |
| VAB-EncodecAudio Tokenizer=Encodec2024.09 | 38.6 | 33.5 | 47.7 | |
| Perceiver2024.03 | 38.4 | 25.8 | 44.2 | |
| Attn AVPretrain=IN SL2024.03 | 38.4 | 25.7 | 46.2 | |
| Perceiver2024.09 | 38.4 | 25.8 | 44.2 | |
| Attn AV2024.09 | 38.4 | 25.7 | 44.2 | |
| VAB-DACAudio Tokenizer=DAC2024.09 | 36.2 | 33.3 | 47 | |
| GBlend2024.03 | 32.4 | 18.8 | 41.8 | |
| G-Blend2024.09 | 32.4 | 18.8 | 41.8 |