Abstract
Recently, methods for action recognition using bone point information have gained much attention and research. However, considering the problem, this paper presents a model of skeletal action recognition based on a dual-residual convolutional neural network. The 3D bone data was used as network input to replace the original network structure with ResNet18 network. With action coherence involved, long and short memory network (LSTM) is used to obtain time features for skeletal movement information flow, further improve the fusion strategy, add static attention mechanism to obtain the most important features, and further extract features. With this model, we achieved 85.3% accuracy on the X-Sub of the NTU RGB+D data set, with higher accuracy compared to the original and other networks.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: 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.
