Action Recognition on HARDVS (test)
0.5963Top-1 AccSMV-EAR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SMV-EARInput Type=Multi-view, Backbone=ResNet18, Data Augmentation=Standard, MACs=1.8G, Params=23.5M2026.01 | 0.5963 | 0.6756 | |
| SMV-EAR*Input Type=Multi-view, Backbone=ResNet18, Data Augmentation=None, MACs=1.8G, Params=23.5M2026.01 | 0.5563 | 0.6356 | |
| MVFNetInput Type=Multi-view, Backbone=ResNet50, MACs=16.4G, Params=23.6M2026.01 | 0.5298 | 0.6326 | |
| TSMInput Type=Frame, Backbone=ResNet50, MACs=0.7G, Params=24.3M2026.01 | 0.5263 | 0.6056 | |
| MVF-NetInput Type=Multi-view, Backbone=ResNet34&18, MACs=2.8G, Params=33.6M2026.01 | 0.5261 | 0.6167 | |
| Swin-TInput Type=Token, Backbone=Transformer, MACs=17.5G, Params=27.8M2026.01 | 0.5191 | 0.5911 | |
| ESTFInput Type=Token, Backbone=ResNet18, MACs=17.6G, Params=46.1M2026.01 | 0.5122 | 0.5753 | |
| TimeSformerInput Type=Token, Backbone=Transformer, MACs=107.3G, Params=121.2M2026.01 | 0.5077 | 0.587 | |
| SlowFastInput Type=Frame, Backbone=ResNet50, MACs=0.7G, Params=33.6M2026.01 | 0.5063 | 0.5777 | |
| CoSTInput Type=Multi-view, Backbone=ResNet50, MACs=9.6G, Params=25.4M2026.01 | 0.5057 | 0.6138 | |
| C3DInput Type=Frame, Backbone=3D CNN, MACs=0.2G, Params=147.2M2026.01 | 0.5052 | 0.5614 | |
| R2Plus1DInput Type=Frame, Backbone=ResNet34, MACs=40.7G, Params=63.5M2026.01 | 0.4906 | 0.5643 | |
| ACTION-NetInput Type=Frame, Backbone=ResNet50, MACs=34.7G, Params=27.9M2026.01 | 0.4685 | 0.5619 | |
| GETInput Type=Token, Backbone=Transformer, MACs=0.9G, Params=4.5M2026.01 | 0.4646 | 0.5237 | |
| ESTInput Type=Learned, Backbone=ResNet34, MACs=2.1G, Params=21.5M2026.01 | 0.3651 | 0.4209 |