A survey on deep neural networks for human action recognition in RGB image and depth image

Human action recognition has been a significant topic in computer vision and video surveillance. With the development of in-depth learning, the application of deep neural networks in related research is increasingly prevalent. This paper provides a survey on deep neural networks for human action recognition, which focus on the methods based on color images or depth images. The detailed description of each method is explained, and several related main datasets are briefly introduced. In this study, all papers published from 2013 to 2015 are reviewed, which provides an overview of the progress in this area.