Audio Recognition on EPIC-KITCHENS 100 (val)
46.05Overall Top-1 Verb AccSlow-Fast (Proposed)
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Slow-Fast (Proposed)# Param.=26.88M2021.03 | 46.05 | 22.95 | 15.22 | 80.01 | 47.98 | 30.24 | 37.56 | 16.34 | 8.83 | |
| Damen et al.# Param.=10.67M2021.03 | 42.63 | 22.35 | 14.48 | 75.84 | 44.6 | 28.23 | 35.4 | 16.34 | 9.2 | |
| Slow-Fast (late fusion)# Param.=25.38M2021.03 | 42.28 | 19.23 | 11.27 | 78.4 | 44.17 | 25.36 | 34.65 | 15.68 | 7.7 | |
| Slow ResNet101# Param.=46.11M2021.03 | 42.24 | 19.35 | 12.12 | 78.14 | 42.83 | 25.3 | 37.37 | 13.9 | 7.61 | |
| Two Slow Streams# Param.=49.78M2021.03 | 41.41 | 19.06 | 11.41 | 77.87 | 43.05 | 24.73 | 34.37 | 14.27 | 6.85 | |
| Slow# Param.=24.89M2021.03 | 41.17 | 18.64 | 11.37 | 77.52 | 42.34 | 24.2 | 34.93 | 14.65 | 7.79 | |
| Fast# Param.=0.49M2021.03 | 39.84 | 17.07 | 8.76 | 76.94 | 41.31 | 22.01 | 33.33 | 15.21 | 6.57 |