Video Action Recognition on Tiny-Kinetics
61.1AccuracyRepresentation Flow
Evaluation Results
| Method | Links | |
|---|---|---|
| Representation FlowIterations=10, Learning flow parameters=true, FcF (Flow-Conv-Flow)=false, Two-stream fusion=false, Resolution=224x224, Number of frames=162018.10 | 61.1 | |
| RGB-OFFIterations=10, Learning flow parameters=true, FcF (Flow-Conv-Flow)=false, Two-stream fusion=false, Resolution=224x224, Number of frames=162018.10 | 55.6 | |
| MFNetIterations=10, Learning flow parameters=true, FcF (Flow-Conv-Flow)=false, Two-stream fusion=false, Resolution=224x224, Number of frames=162018.10 | 52.5 | |
| ActionFlownetIterations=10, Learning flow parameters=true, FcF (Flow-Conv-Flow)=false, Two-stream fusion=false, Resolution=224x224, Number of frames=162018.10 | 51.8 | |
| TVNetIterations=10, Learning flow parameters=true, FcF (Flow-Conv-Flow)=false, Two-stream fusion=false, Resolution=224x224, Number of frames=162018.10 | 39.4 |