Abstract
Continuous human action recognition (CHAR) is more practical in human-robot interactions. In this paper, an online CHAR algorithm is proposed based on skeletal data extracted from RGB-D images captured by Kinect sensors. Each human action is modeled by a sequence of key poses and atomic motions in a particular order. In order to extract key poses and atomic motions, feature sequences are divided into pose feature segments and motion feature segments, by use of the online segmentation method based on potential differences of features. Likelihood probabilities that each feature segment can be labeled as the extracted key poses or atomic motions, are computed in the online model matching process. An online classification method with variable-length maximal entropy Markov model (MEMM) is performed based on the likelihood probabilities, for recognizing continuous human actions. The variable-length MEMM method ensures the effectiveness and efficiency of the proposed CHAR method. Compared with the published CHAR methods, the proposed algorithm does not need to detect the start and end points of each human action in advance. The experimental results on public datasets show that the proposed algorithm is effective and highly-efficient for recognizing continuous human actions.
Highlights
Human action recognition is a crucial and challenging task in many research and application fields, such as video surveillance [1,2], human-robot interactions [3]
Human skeletal data can be extracted from RGB-D data which are captured by a Kinect sensor [6], and human actions can be modeled as a continuous evolution of human skeletal joints [7]
This demonstrates that the proposed variable-length maximal entropy Markov model (MEMM) method can take full use of different roles of discriminatory and neutral key poses or atomic motions in human action recognition, and can utilize temporal sequence information among key poses and atomic motions more effectively
Summary
Human action recognition is a crucial and challenging task in many research and application fields, such as video surveillance [1,2], human-robot interactions [3]. Most of the published human action recognition methods up to now mainly focus on segmented and unified action classification, i.e., to identify the category of each data sequence which only contains one single human action performed by one single person [8,9,10,11]. Public datasets, such as MSRAction3D [12], MSRC-12 [13], and CAD-60 [14], supply data sequences which have been segmented according to action categories and performers. One RGB-D data sequence may contain several kinds of human actions which are not segmented in advance, Sensors 2016, 16, 161; doi:10.3390/s16020161 www.mdpi.com/journal/sensors so it may to each detect the start and end frames human if wemethods want to to apply start and be endnecessary frames of human action, if we want of toeach apply some action, published the some published methods to the practical human daily life
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.