Human activity recognition with semantically guided graph-convolutional network

Recognizing specific actions, activities and goals of an individual in any environment (constraint or unconstrained) will be a key feature for holographic communications in order to achieve a sense of reality and naturalness of the face-to-face interaction. While activity and action recognition are tasks that humans perform naturally and with little exertion, they are still a challenge from the point of view of artificial intelligence in the context of deep learning. The main goal of this paper is to present a technique for human activity recognition with semantically guided graph-convolutional network based on auto-regressive moving average (ARMA) filters, for the purpose of holographic communication. The semantic is introduced by the human skeleton representation. By recognizing the activity, we can plan for the next step in the proposed architecture: prediction; thus solving some of the challenges imposed by real time constraints and channel limits when transmitting huge and heterogeneous amounts of data for this type of communication even in 5G.

• Agata Manolova
• Krasimir Tonchev
• Nikolay Neshov
• Nicole Christoff

2021 XXX International Scientific Conference Electronics (ET)

https://ieeexplore.ieee.org/document/9580051