Human Activity Recognition on Smartphone Using Transformer Network
Sachin Singh, Arjun Singh, Vikrant Singh Rana, Shobhit Prajapati, Saurabh Bhardwaj, Pawan Kr. Pal · 2023
Automatic behaviour inspection for sports players, older people, and applications for 1OT require sensor-based human activity identification. Conventional means rely on finger features to tend to fluctuate from raw data using preset mathematical concepts and therefore are incapable of domain adaptation. We proposed a Transformer network in this study to recognize six human actions from Smartphone data. Each participant did several activities while holding a Mobile device around their waist (moving, going up, and coming down, resting, constant, and set down). We first should define the Titans model using the Tensor Flow backend module the tests were recorded and labelled manually and partitioned randomly into two groups. Furthermore, the activity of the project has been examined.