Action Recognition on HMDB-51 (test)
0.1Total Time (ms)DMC-Net
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DMC-NetScenario=Generator Only, Batch Size=12019.01 | 0.1 | — | — | — | 9,433.9 | |
| DMC-NetScenario=Generator + Classifier, Backbone=ResNet-18, Batch Size=12019.01 | 0.3 | — | — | — | 3,333.3 | |
| CoViARStream=Compressed Video Based, Forwarding Scenario=Concurrent (C), Batch Size=12019.01 | 0.68 | 0.46 | 0.22 | 4,545.4 | 1,470.5 | |
| DMC-NetStream=Compressed Video Based, Forwarding Scenario=Concurrent (C), Batch Size=12019.01 | 0.76 | 0.46 | 0.3 | 3,333.3 | 1,315.7 | |
| CoViARStream=Compressed Video Based, Forwarding Scenario=Sequential (S), Batch Size=12019.01 | 1.05 | 0.46 | 0.59 | — | — | |
| DMC-NetStream=Compressed Video Based, Forwarding Scenario=Sequential (S), Batch Size=12019.01 | 1.35 | 0.46 | 0.89 | — | — | |
| PWC-NetScenario=Generator Only, Batch Size=12019.01 | 28.6 | — | — | — | 35 | |
| PWC-NetScenario=Generator + Classifier, Backbone=ResNet-18, Batch Size=12019.01 | 28.8 | — | — | — | 34.8 | |
| BN-InceptionStream=Two-Stream (RGB+Flow), Forwarding Scenario=Concurrent (C), Batch Size=12019.01 | 75.9 | 75 | 0.9 | 1,111.1 | 13.1 | |
| BN-InceptionStream=Two-Stream (RGB+Flow), Forwarding Scenario=Sequential (S), Batch Size=12019.01 | 76.6 | 75 | 1.6 | — | — | |
| ResNet152Stream=Two-Stream (RGB+Flow), Forwarding Scenario=Concurrent (C), Batch Size=12019.01 | 79 | 75 | 4 | 250 | 12.6 | |
| ResNet152Stream=Two-Stream (RGB+Flow), Forwarding Scenario=Sequential (S), Batch Size=12019.01 | 82.5 | 75 | 7.5 | — | — | |
| TVNetScenario=Generator Only, Batch Size=12019.01 | 83.3 | — | — | — | 12 | |
| TVNetScenario=Generator + Classifier, Backbone=ResNet-18, Batch Size=12019.01 | 83.5 | — | — | — | 12 | |
| Flownet2.0Scenario=Generator Only, Batch Size=12019.01 | 220.8 | — | — | — | 4.5 | |
| Flownet2.0Scenario=Generator + Classifier, Backbone=ResNet-18, Batch Size=12019.01 | 221 | — | — | — | 4.5 | |
| DeepflowScenario=Generator Only, Batch Size=12019.01 | 1,449.2 | — | — | — | 0.7 | |
| DeepflowScenario=Generator + Classifier, Backbone=ResNet-18, Batch Size=12019.01 | 1,449.5 | — | — | — | 0.7 |