论文部分内容阅读
This research presents a novel way of labelling human activities from the skeleton output computed from RGB-D data from vision-based motion capture systems. The activities are labelled by means of a Compound Hidden Markov Model. The linkage of several Lin