Audio Tagging on Full AudioSet 1.0 (test)
47.44mAPProposed Full Model (PSLA Ensemble)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Proposed Full Model (PSLA Ensemble)#Params=13.6M x 102021.02 | 47.44 | 98.1 | 2.936 | |
| Proposed 68M Model (PSLA Ensemble)#Params=13.6M x 52021.02 | 46.9 | 97.89 | 2.872 | |
| Proposed Single Model (PSLA)#Params=13.6M2021.02 | 44.35 | 97.53 | 2.778 | |
| PANNs#Params=81M2021.02 | 43.9 | 97.3 | 2.725 | |
| DeepRes#Params=26M2021.02 | 39.2 | 97.1 | 2.682 | |
| Baseline (EfficientNet)#Params=13.6M2021.02 | 37.23 | 97.06 | 2.672 | |
| Kongyear=20192021.02 | 36.9 | 96.9 | 2.639 | |
| TALNet2021.02 | 36.2 | 96.5 | 2.554 | |
| Yu2021.02 | 36 | 97 | 2.66 | |
| Kongyear=20182021.02 | 32.7 | 96.5 | 2.558 | |
| AudioSet Baseline2021.02 | 31.4 | 95.9 | 2.452 |