based motion recognition for smart fitness application
Recently, a lot of applications have been developed using motion recognition technology, such as home training and unmanned shop. In this research, we are developing a novel motion recognition algorithm based on the mono image data based on GNN (Graph Neural Network)
< based motion recognition for smart fitness application >
We are developing a stable and accurate motion recognition algorithm based on GNN (Graph Neural Network). The GNN inputs the skeleton data extracted from the images and outputs the motion estimation. Our algorithm focused on improving the estimation accuracy and the computational efficiency. We have built our own motion datasets for Takowndo actions and developing the GNN-based motion recognition algorithm using those datasets.
< Sensor fusion for outdoor localization >