Spatial–temporal graph neural network based on node attention
Publié en ligne: 08 avr. 2022
Pages: 703 - 712
Reçu: 23 mai 2021
Accepté: 27 sept. 2021
© 2021 Qiang Li et al., published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.
Fig. 1
(a) Spatial–temporal skeletal graph. (b) Partitioning strategy graph.Fig. 2
Structure diagram of NA-STGCN. GCN, graph convolutional neural network; NA-STGCN, node attention spatial–temporal graph convolutional neural network.Fig. 3
Change curve of loss values. NA-STGCN, node attention spatial–temporal graph convolutional neural network; ST-GCN, spatial–temporal graph convolutional neural network.Fig. 4
Node activation response maps: (a) top left: ST-GCN clapping action; (b) top right: NA-STGCN clapping action; (c) bottom left: ST-GCN brushing action; (d) bottom right: NA-STGCN brushing action. NA-STGCN, node attention spatial–temporal graph convolutional neural network; ST-GCN, spatial–temporal graph convolutional neural network.Comparison with representative methods (%).
Model |
CS |
CV |
Two-Stream 3DCNN [8] |
66.8 |
72.6 |
TCN [22] |
74.3 |
83.1 |
Clip + CNN + MTLN [23] |
79.6 |
84.8 |
VA-LSTM [10] |
79.4 |
87.6 |
ST-GCN [13] |
81.5 |
88.3 |
NA-STGCN (ours) |
85.8 |
89.3 |