Self Supervised Human Action Recognition On
評価指標
Classifier
Encoder
xset (%)
xsub (%)
評価結果
このベンチマークにおける各モデルのパフォーマンス結果
比較表
| モデル名 | Classifier | Encoder | xset (%) | xsub (%) |
|---|---|---|---|---|
| augmented-skeleton-based-contrastive-action | FC | LSTM | 49.2 | 48.6 |
| contrastive-learning-from-extremely-augmented | FC | ST-GCN | 68.8 | 68.2 |
| modeling-the-relative-visual-tempo-for-self | FC | ST-GCN | 68.9 | 68.0 |
| prototypical-contrast-and-reverse-prediction | FC | GRU | 45.1 | 41.7 |
| predict-cluster-unsupervised-skeleton-based | FC | GRU | 42.7 | 41.7 |
| unsupervised-motion-representation-learning | FC | MCAE | 54.7 | 52.8 |
| cross-model-cross-stream-learning-for-self | FC | ST-GCN | 71.7 | 68.5 |
| spatiotemporal-decouple-and-squeeze | FC | DSTA | 55.6 | 50.6 |