CVNov 10, 2024
Extended multi-stream temporal-attention module for skeleton-based human action recognition (HAR)
arXiv:2411.06553v119 citationsh-index: 10Comput Hum Behav
Originality Synthesis-oriented
AI Analysis
This is an incremental improvement for researchers in human action recognition, addressing specific limitations in GCN-based models.
The paper tackled the problem of rigid graph structures in graph convolutional networks (GCNs) for skeleton-based human action recognition by proposing an extended multi-stream temporal-attention module, achieving a 2.5% accuracy improvement on the NTU RGB+D dataset.
Graph convolutional networks (GCNs) are an effective skeleton-based human action recognition (HAR) technique. GCNs enable the specification of CNNs to a non-Euclidean frame that is more flexible. The previous GCN-based models still have a lot of issues: (I) The graph structure is the same for all model layers and input data.