Table 4.
Comparison of spatial graph convolution layer with and without SE block on the Filtered NTU RGB+D dataset.
| Methods | Accuracy (%) |
|---|---|
| J-ASGCN w/o SE | 93.6 |
| B-ASGCN w/o SE | 93.1 |
| ASGCN w/o SE | 94.3 |
| J-ASGCN | 94.0 |
| B-ASGCN | 93.2 |
| ASGCN | 94.9 |
The bold values indicate the accuracy of the model incorporating the attention mechanism.