Video Recognition on FCVID (test)
84.4mAPUpper
Evaluation Results
| Method | Links | |
|---|---|---|
| UpperInput frames=16, GFLOPs=117.3, #3D=5.0, #Frame=16.02020.12 | 84.4 | |
| Ada3DInput frames=16, GFLOPs=66.6, #3D=2.1, #Frame=10.72020.12 | 84.3 | |
| NSNetBackbone=ResNet-50, GFLOPS=26.0G2022.07 | 83.9 | |
| TSQNetFLOPS=26.2G2022.07 | 83.5 | |
| AdaFocusBackbone=ResNet-50, GFLOPS=26.6G2022.07 | 83.4 | |
| AdaFocusFLOPS=26.6G2022.07 | 83.4 | |
| TSQNetVQM-only=true, Resolution=224, FLOPS=26.2G2022.07 | 83.3 | |
| TSQNetVQM-only=true, Resolution=188, FLOPS=24.4G2022.07 | 82.9 | |
| VideoIQBackbone=ResNet-50, GFLOPS=27.0G2022.07 | 82.7 | |
| VideoIQFLOPS=27.0G2022.07 | 82.7 | |
| UpperInput frames=8, GFLOPs=58.6, #3D=5.0, #Frame=8.02020.12 | 82.1 | |
| SMARTBackbone=ResNet-502022.07 | 82.1 | |
| SMART2022.07 | 82.1 | |
| Random FTInput frames=16, GFLOPs=65.3, #3D=2.1, #Frame=10.62020.12 | 82 | |
| Ada3DInput frames=8, GFLOPs=35.6, #3D=2.2, #Frame=5.72020.12 | 81.9 | |
| AdaFuseBackbone=ResNet-50, GFLOPS=45.0G2022.07 | 81.6 | |
| AdaFuseFLOPS=45.0G2022.07 | 81.6 | |
| AR-NetBackbone=ResNet-50, GFLOPS=35.1G2022.07 | 81.3 | |
| AR-NetFLOPS=35.1G2022.07 | 81.3 | |
| SCSamplerBackbone=ResNet-50, GFLOPS=42.0G2022.07 | 81 | |
| SCSamplerFLOPS=42.0G2022.07 | 81 | |
| Random FTInput frames=8, GFLOPs=36.1, #3D=2.2, #Frame=5.82020.12 | 80.7 | |
| AdaFrameBackbone=ResNet-50, GFLOPS=75.1G2022.07 | 80.2 | |
| AdaFrameFLOPS=75.1G2022.07 | 80.2 | |
| LiteEvalBackbone=ResNet-50, GFLOPS=94.3G2022.07 | 80 | |
| LiteEvalFLOPS=94.3G2022.07 | 80 | |
| RandomInput frames=16, GFLOPs=63.2, #3D=2.1, #Frame=10.32020.12 | 79.2 | |
| RandomInput frames=8, GFLOPs=36.1, #3D=2.2, #Frame=5.82020.12 | 78.1 |