A Semi-supervised Human Action Recognition Algorithm Based on Skeleton Feature

@inproceedings{Yuan2014ASH,
  title={A Semi-supervised Human Action Recognition Algorithm Based on Skeleton Feature},
  author={Hejin Yuan},
  year={2014}
}
A semi-supervised human action recognition algorithm using skeleton feature is put forward in this paper. In the method, the cumulative skeleton image and skeleton history image are firstly calculated as the feature representation of the human actions. Then, the label of unannotated actions is predicted through the constrained semisupervised K-means clustering algorithm. Meanwhile the average cumulative and history skeleton images are generated as the model of each category actions. Finally… CONTINUE READING