Boosting VLAD with double assignment using deep features for action recognition in videos | IEEE Conference Publication | IEEE Xplore