Action Recognition on HMDB51 (40% labels)
42.1AccuracyL2A
Evaluation results
| Method | Links | |
|---|---|---|
| L2ABackbone=3D ResNet-18, Selector Pre-training=true2022.06 | 42.1 | |
| L2ABackbone=3D ResNet-18, Selector Pre-training=false2022.06 | 37.9 | |
| ActorCut+IDConference=Arxiv21, Backbone=3D ResNet-182022.06 | 35.7 | |
| TCLConference=ICCV21, Backbone=3D ResNet-182022.06 | 34.8 | |
| ActorCutConference=Arxiv21, Backbone=3D ResNet-182022.06 | 32.9 | |
| VideoSSLConference=WACV21, Backbone=3D ResNet-182022.06 | 32.7 | |
| S4LConference=ICCV19, Backbone=3D ResNet-182022.06 | 29.8 | |
| ST-VideoMixConference=Arxiv21, Backbone=3D ResNet-182022.06 | 28.7 | |
| CutMixConference=ICCV19, Backbone=3D ResNet-182022.06 | 27.8 | |
| MixUpConference=ICLR18, Backbone=3D ResNet-182022.06 | 27.5 | |
| PseudoLabelConference=ICMLW13, Backbone=3D ResNet-182022.06 | 27.3 | |
| MeanTeacherConference=Neurips17, Backbone=3D ResNet-182022.06 | 27.2 | |
| CutOutConference=Arxiv17, Backbone=3D ResNet-182022.06 | 27.1 |