Joint visual-temporal embedding for unsupervised learning of actions in untrimmed sequencesRosaura G. VidalmataWalter J. Scheireret al.2021WACV 2021