An expressive deep model for human action parsing from a single image | IEEE Conference Publication | IEEE Xplore