Combining Contextual and Modal Action Information into a Weighted Multikernel SVM for Human Action Recognition | Publicación