Construct Dynamic Graphs for Hand Gesture Recognition via Spatial-Temporal Attention
This work addresses robust hand gesture recognition for applications like human-computer interaction, but it is incremental as it builds on existing graph and attention methods.
The paper tackles hand gesture recognition by proposing a dynamic graph-based spatial-temporal attention method that constructs graphs from hand skeletons and uses attention mechanisms, achieving superior performance on benchmarks like DHG-14/28 and SHREC'17 while reducing computational cost by 99%.
We propose a Dynamic Graph-Based Spatial-Temporal Attention (DG-STA) method for hand gesture recognition. The key idea is to first construct a fully-connected graph from a hand skeleton, where the node features and edges are then automatically learned via a self-attention mechanism that performs in both spatial and temporal domains. We further propose to leverage the spatial-temporal cues of joint positions to guarantee robust recognition in challenging conditions. In addition, a novel spatial-temporal mask is applied to significantly cut down the computational cost by 99%. We carry out extensive experiments on benchmarks (DHG-14/28 and SHREC'17) and prove the superior performance of our method compared with the state-of-the-art methods. The source code can be found at https://github.com/yuxiaochen1103/DG-STA.