Selective spatiotemporal features learning for dynamic gesture recognition

作者：

Highlights：

• 3D convolutional neural networks can learn spatial–temporal features directly.

• Convolutional LSTM is suitable to encode long-term temporal information.

• The parallel combination of various networks makes full use of their advantages.

• The no-linear aggregation approach empowers neurons adaptive.

摘要

•3D convolutional neural networks can learn spatial–temporal features directly.•Convolutional LSTM is suitable to encode long-term temporal information.•The parallel combination of various networks makes full use of their advantages.•The no-linear aggregation approach empowers neurons adaptive.

论文关键词：Dynamic gesture recognition,Deep learning,Spatiotemporal features learning,Heterogeneous network,Attention mechanism

论文评审过程：Received 4 July 2020, Revised 20 November 2020, Accepted 16 December 2020, Available online 5 January 2021, Version of Record 7 January 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114499