Enhanced skeleton visualization for view invariant human action recognition

作者：

Highlights：

• Sequence-based view invariant transform can effectively cope with view variations.

• Enhanced skeleton visualization method encodes spatio-temporal skeletons as visual and motion enhanced color images in a compact yet distinctive manner.

• Multi-stream convolutional neural networks fusion model is able to explore complementary properties among different types of enhanced color images.

• Our method consistently achieves the highest accuracies on four datasets, including the largest and most challenging NTU RGB+D dataset for skeleton-based action recognition.

摘要

•Sequence-based view invariant transform can effectively cope with view variations.•Enhanced skeleton visualization method encodes spatio-temporal skeletons as visual and motion enhanced color images in a compact yet distinctive manner.•Multi-stream convolutional neural networks fusion model is able to explore complementary properties among different types of enhanced color images.•Our method consistently achieves the highest accuracies on four datasets, including the largest and most challenging NTU RGB+D dataset for skeleton-based action recognition.

论文关键词：Human action recognition,View invariant,Skeleton sequence

论文评审过程：Received 14 December 2016, Revised 18 February 2017, Accepted 25 February 2017, Available online 3 March 2017, Version of Record 22 April 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.02.030