Using Action Objects Contextual Information for a Multichannel SVM in an Action Recognition Approach based on Bag of VisualWords | Publicación