Weakly supervised learning of actions from transcripts

Hilde Kuehne, Alexander Richard, Juergen Gall

URL

Proceedings: CVIU

Year: 2017

Peer Reviewed : Yes

Results: 
Hollywood Extended /HE Action seg. weakly (MoF): 33% (GMM)
Hollywood Extended /HE Action alignment weakly (IoD): 46% (GMM)
Breakfast Dataset /BF Action seg. weakly (MoF): 28.2% (GMM + CNN )
Breakfast Dataset /BF Action alignment weakly (IoD): 46% (GMM + CNN )