Dynamic hand gesture recognition has attracted increasing interests becauseof its importance for human computer interaction. In this paper, we propose anew motion feature augmented recurrent neural network for skeleton-baseddynamic hand gesture recognition. Finger motion features are extracted todescribe finger movements and global motion features are utilized to representthe global movement of hand skeleton. These motion features are then fed into abidirectional recurrent neural network (RNN) along with the skeleton sequence,which can augment the motion features for RNN and improve the classificationperformance. Experiments demonstrate that our proposed method is effective andoutperforms start-of-the-art methods.
展开▼