Learning content and style: Joint action recognition and person identification from human skeletons

作者：

Highlights：

• We are the first to pair action recognition and person identification to imitate the ability of our visual system.

• We propose a new end-to-end trainable pipeline, which consists of skeleton transformation and multi-task RNN.

• We propose several novel architectures of multi-task RNN with different amounts of sharing layers.

• Experiments show that for these two tasks, learning one task would benefit from learning another task.

摘要

•We are the first to pair action recognition and person identification to imitate the ability of our visual system.•We propose a new end-to-end trainable pipeline, which consists of skeleton transformation and multi-task RNN.•We propose several novel architectures of multi-task RNN with different amounts of sharing layers.•Experiments show that for these two tasks, learning one task would benefit from learning another task.

论文关键词：Content and style,Action recognition,Person identification from motions,Skeleton transformation,Multi-task RNN

论文评审过程：Received 15 May 2017, Revised 20 March 2018, Accepted 27 March 2018, Available online 27 March 2018, Version of Record 4 April 2018.

论文官网地址：https://doi.org/10.1016/j.patcog.2018.03.030